34
00:02:32.420 --> 00:02:57.660
Anuradha Goli(Anu): Great. Yeah, I think I think we have a good few people in here. Okay, we're gonna get started. Hi, good morning. Good afternoon, and good evening, everybody I am, Anu. I'm actually co-hosting this session with Mohan. I'm very excited to have Andrew with us today. So

35
00:02:58.050 --> 00:03:05.319
Anuradha Goli(Anu): it's he like just to give you a bit of intro about Andrew. He has more than 22 years of experience.

36
00:03:06.270 --> 00:03:13.020
Anuradha Goli(Anu): As an engineering leader and a product leader. And so through his current company, Eden soft labs.

37
00:03:13.220 --> 00:03:20.370
Anuradha Goli(Anu): he would like to share his expertise with external organizations to achieve the business agility. So

38
00:03:20.640 --> 00:03:24.249
Anuradha Goli(Anu): and over to you, Andrew, before before I

39
00:03:24.480 --> 00:03:40.129
Anuradha Goli(Anu): give to you just guys, if you have any questions, please post them in the chat. Myself and Mohan will look after those questions, and we will have a Q&A at the end, and so we will. We will take all those questions at the end, and so over to you, Andrew.

40
00:03:41.400 --> 00:03:42.450
Andrew Park: All right.

41
00:03:44.410 --> 00:03:46.470
Andrew Park: Okay, can everybody see my screen?

42
00:03:48.570 --> 00:04:09.690
Andrew Park: Alright? So today, I'm hoping to give you a roadmap to greater productivity and skill growth. I'm going to talk a lot about the impact of Gen. AI, and how to infuse it into your job. Or if you're a leader, how to have a vision to infuse it into your organization.

43
00:04:14.660 --> 00:04:19.152
Andrew Park: Okay, let's see how come. I can't get to the next slide.

44
00:04:19.680 --> 00:04:32.399
Andrew Park: Oh, there it goes. Okay. So I'm the founder of Eden soft labs consulting which is a consulting arm within G 3 technologies, this place that I've been working at forever.

45
00:04:32.590 --> 00:04:40.820
Andrew Park: And I'm an active collaborator with agile alliances, reimagining agile initiative, headed up by David Luke.

46
00:04:41.481 --> 00:04:55.600
Andrew Park: Some background in the last 18 months I've probably had conversations with more than 50 Ceos. I also have been doing in depth interviews with a lot of product managers at a lot of different companies.

47
00:04:55.750 --> 00:05:06.159
Andrew Park: And I've been passing these insights onto the agile Alliance in order for them to be able to have data to understand what's happening in the industry.

48
00:05:08.580 --> 00:05:26.380
Andrew Park: something about me. I'm an experienced product manager, design leader and engineering leader. I was able to wear all these hats because I was there when we were a startup. And when when you're working for a startup, you actually get the opportunity to wear multiple hats.

49
00:05:27.990 --> 00:05:36.320
Andrew Park: more importantly, for this talk, I've spearheaded the adoption of machine learning and AI into my teams and products during the last decade.

50
00:05:36.827 --> 00:05:41.310
Andrew Park: We have machine learning or AI in most of our products.

51
00:05:41.771 --> 00:05:54.439
Andrew Park: We also, have a self hosted open source Gen. AI using open source Llms internally, so that we can leverage that without having to share all of our data with big tech

52
00:05:54.889 --> 00:06:14.370
Andrew Park: but we also use lots of vendor AI tools extensively. I think that most of you probably don't have the luxury of being able to have open source. Gen. AI running inside your walls. So you're probably going to be using Chat Gpt, or something like that. So I'm going to be talking a lot in that context of how to use chat Gpt.

53
00:06:14.860 --> 00:06:23.250
Andrew Park: and I'm an avid proponent of every staff member aggressively using Gen. AI all day every day.

54
00:06:23.871 --> 00:06:49.200
Andrew Park: I've been saying this to my own staff for the last 18 months, and and this is something that when Chat Gpt 3.5 came out. It was just amazing. I I just really ramped up the leadership to be able to make sure that everybody understood that they needed to figure out how to use this in their company or in their responsibilities.

55
00:06:50.090 --> 00:06:55.479
Andrew Park: So this is a slide that I showed to my software developers.

56
00:06:56.140 --> 00:07:01.399
Andrew Park: Really, probably just a few months after Chat, Gpt, 3.5 came out.

57
00:07:01.550 --> 00:07:08.909
Andrew Park: And I talked with them by starting, saying, Okay, the world is changing.

58
00:07:09.240 --> 00:07:16.080
Andrew Park: And because of chat Gpt, I'm not sure that we need to hire software developers anymore.

59
00:07:16.410 --> 00:07:20.740
Andrew Park: And then I saw puzzle looks because we're a software development company.

60
00:07:21.000 --> 00:07:24.989
Andrew Park: And then I said, and I'm not sure a year from now.

61
00:07:25.120 --> 00:07:27.760
Andrew Park: if we have jobs for software developers.

62
00:07:27.950 --> 00:07:32.450
Andrew Park: And then I saw looks of fear on faces. All right. Now I had their attention.

63
00:07:32.890 --> 00:07:40.800
Andrew Park: and I said, what we do have jobs for a year from now is for software composers. And let me tell you what the difference is

64
00:07:41.486 --> 00:07:47.870
Andrew Park: a software composer doesn't spend all their time doing code implementation and unit testing.

65
00:07:48.363 --> 00:07:53.869
Andrew Park: They use AI for that, because AI can be really good at writing methods.

66
00:07:54.120 --> 00:08:14.030
Andrew Park: But what you do need to do is get really good at design, because AI is not very good at architectural stuff, it won't even create a good class ecosystem for you. You need how to know how to do that really well, and then leverage AI to do all the methods

67
00:08:14.300 --> 00:08:32.800
Andrew Park: and then leverage AI to do unit tests and so forth. So basically, I was trying to paint a picture to them that they need to operate at a higher level. I was painting a picture to them that I expected them to adopt this technology very aggressively.

68
00:08:33.240 --> 00:08:51.790
Andrew Park: and the months after that I would check in with developers every now and then. Say so. What you've been doing this week. Oh, I was writing code for this writing code for that. Why are you writing code? Chat Gpt should be doing that for you. You should just be composing it into something larger, and editing it to make it maintainable.

69
00:08:52.110 --> 00:08:53.310
Andrew Park: And so

70
00:08:53.620 --> 00:09:09.059
Andrew Park: part of the leadership that I gave was really breaking people out of the habits that they had of doing things themselves when they could be using Chat Gpt or AI to be able to greatly speed up those tasks.

71
00:09:09.390 --> 00:09:16.970
Andrew Park: Now, what you're looking at. Here is a mind map of the different tasks that a software developer will carry out.

72
00:09:17.610 --> 00:09:26.779
Andrew Park: And what's in red is what Gen. AI can highly displace, or you could say, could highly accelerate.

73
00:09:26.970 --> 00:09:48.959
Andrew Park: What's in yellow are the things that Gen. AI can moderately displace, or you could say, moderately accelerate. And what's in green are things that largely the human still has to do. So you can see from this that just from this mind map that Gen. AI can and should have a pretty big impact on a software developer's job.

74
00:09:49.110 --> 00:10:11.440
Andrew Park: Now, during this presentation, I'm going to get to several other roles, so that if you're a scrum, master, you can see how AI affects you. If you're a product owner, you can see how scrum, master, how how AI affects you, or designer and so I'm going to get into a lot of details. And I'm going to give you practical tips of things to try out.

75
00:10:13.490 --> 00:10:31.869
Andrew Park: Okay. So one of the things in the bigger picture that I want to talk about is that software delivery is undergoing evolutionary changes right now. And this is due to the explosion of complexity. In software programs as well as elevated Ux requirements.

76
00:10:32.050 --> 00:10:42.169
Andrew Park: And so what's happening is that the rise of AI, coupled with the rise of the product management field

77
00:10:42.280 --> 00:10:57.070
Andrew Park: are combining together where I'm seeing evidence that leading companies, particularly in Silicon Valley, have been really changing their software delivery practices and to accomplish what I'm calling product team agility.

78
00:10:57.240 --> 00:11:11.779
Andrew Park: So instead of focusing on the agility of the delivery teams of the of the agile teams, it's Deli. It's agility at the larger level, including product management and designers.

79
00:11:12.580 --> 00:11:18.089
Andrew Park: Now, one of the things that I'm seeing here

80
00:11:18.210 --> 00:11:31.520
Andrew Park: is that this trend it's being pushed by Ceos wanting to launch whole new products more often. So this is a big change from 20 years ago, when agile and scrum came on the scene.

81
00:11:31.730 --> 00:11:48.540
Andrew Park: And this is because product lifespans are shorter. If you're paying attention to product lifespans over time, you may have noticed this. I guarantee you, Ceos, know this. And so what I'm finding is that Ceos

82
00:11:48.730 --> 00:12:06.490
Andrew Park: don't really want to hear or talk about agile anymore. They don't really want to hear or talk about scrum anymore. It's not exciting to them. They know what that is. They've been there done that what? But what they're very interested in is business agility. So that's something that's like a super hot term. And this is the reason why

83
00:12:06.830 --> 00:12:13.879
Andrew Park: they need to launch new products. More often they want they trying to deal with the fact that product lifespans are shorter.

84
00:12:14.080 --> 00:12:16.650
Andrew Park: And so what they want.

85
00:12:16.770 --> 00:12:38.350
Andrew Park: our frameworks, practices, and tools that provide more business agility than what agile and scrum have been providing this last 20 years. So I've talked with a lot of companies where they feel like a 2 week sprint is like way too long, you know, things like that. And so sometimes they're like, Yeah, why can't it be a day? Why, why do I have to wait 2 weeks?

86
00:12:38.480 --> 00:12:42.680
Andrew Park: So there's big big changes happening. And this is the reason why.

87
00:12:43.000 --> 00:12:43.960
Andrew Park: Now

88
00:12:44.130 --> 00:12:51.139
Andrew Park: some of you might have heard the big news of the strategic partnership between agile Alliance and the Pmi.

89
00:12:51.720 --> 00:12:57.639
Andrew Park: This, these factors here are one of the things driving that

90
00:12:57.940 --> 00:13:21.180
Andrew Park: so I've seen a lot of posts where people have been fearful saying like, I don't understand this. It's like the Rebel Alliance joining the Empire, you know, because they're still teaching waterfall. And and they're missing the bigger picture. The bigger picture is that the Pmi has a lot of respect and attention from the sea level executives.

91
00:13:21.350 --> 00:13:39.289
Andrew Park: and they know that the world is changing. We know that the world is changing and within agile alliance, and we're trying to join forces so that we can help each other with this massive change that you're going to see happen in this next couple of years.

92
00:13:39.480 --> 00:13:52.549
Andrew Park: So it's a really good time to pay attention to what's happening through these agile tech talks within agile alliance because you're gonna see a lot of new stuff being presented.

93
00:13:54.910 --> 00:13:59.269
Andrew Park: Okay, so let's talk about product team agility.

94
00:13:59.450 --> 00:14:08.430
Andrew Park: So the rules are changing from what I'm seeing, and I want to give you advice so that you can ensure that you're part of the future.

95
00:14:08.790 --> 00:14:17.520
Andrew Park: So some of my observations from speaking to lots of Ceos and product managers are this. They are evolving their team structures.

96
00:14:17.650 --> 00:14:46.780
Andrew Park: They're doing more with less. So everyone's aware of the layoffs and teams are shrinking in size. And now people are expected to become cross functional performers. And so this is replacing the rigid roles. And that's what I'm showing here is that over here. Maybe you had a team of 5, and now that becomes a team of 3, and but they're more widely skilled because they're being forced to do that.

97
00:14:47.511 --> 00:14:58.339
Andrew Park: I'm finding lots of examples where where scrum masters have been. Let go. Those responsibilities are now being absorbed by product managers

98
00:14:58.440 --> 00:15:24.159
Andrew Park: in situations where product owners have been let go. Those responsibilities have been absorbed by certain engineers. Often the tech leads, and that has given rise to a new job description in Silicon Valley called Product Engineer. It's basically one of the engineers that also has taken on the product ownership role. And now they have to make sure that the rest of the team understands and is aligned.

99
00:15:24.200 --> 00:15:30.319
Andrew Park: So these cost cutting measures are having a big effect on team structure.

100
00:15:30.540 --> 00:15:35.680
Andrew Park: So my advice to you, based on these industry trends number one

101
00:15:35.910 --> 00:15:39.170
Andrew Park: ally and align yourself with product and sales.

102
00:15:39.290 --> 00:16:00.410
Andrew Park: So if you have been spending all your time with agile team and not much time with your product managers. Not much time with your salespeople. My advice to you is to change that immediately. Try to make friends, ally, align yourself, understand what their needs are, where they want the business to go.

103
00:16:01.110 --> 00:16:15.726
Andrew Park: Second thing is focus on business outcomes and not metrics. This is something that I think will really endear you toward people in product and sales, because

104
00:16:16.680 --> 00:16:39.969
Andrew Park: story points. Feature counts, Dora metrics. All of these things are not the same thing as guaranteeing business outcomes. They care about business outcomes, and so try to understand that and pay less attention to the metrics. The 3rd thing is to leverage AI to amplify your impact. And that's what the rest of this talk is all about.

105
00:16:44.190 --> 00:16:53.850
Andrew Park: Okay, so this is where this journey is gonna take you part one. I'm gonna talk about the impact of Gen. AI,

106
00:16:54.260 --> 00:16:56.850
Andrew Park: and there's good news. And there's bad news.

107
00:16:57.370 --> 00:16:58.540
Andrew Park: Part 2

108
00:16:58.640 --> 00:17:21.639
Andrew Park: is, I'm going to talk a little bit about Gen. AI strengths and limitations. I'm not really gonna talk about all the aspects of AI or anything like that this is not going to be a highly highly technical talk. I'm going to try to say this so that at a layman's perspective you understand what you need to understand, to know how to leverage it and how to get the most out of it.

109
00:17:22.260 --> 00:17:32.710
Andrew Park: And part 3 is that there's a lot of content. I have on AI tools that you could look into to streamline your workflows and amplify your talent.

110
00:17:35.320 --> 00:17:41.829
Andrew Park: Okay? So 1st part, the impact of Gen. AI on the workplace. There's good news and bad news.

111
00:17:42.390 --> 00:17:50.130
Andrew Park: The bad news is that Gen. AI is displacing jobs on our Youtube channel.

112
00:17:50.805 --> 00:17:55.460
Andrew Park: I posted an interview with this civil engineering business owner.

113
00:17:55.640 --> 00:18:02.919
Andrew Park: And it's a story where he had employed 2 and a half software developers.

114
00:18:03.450 --> 00:18:07.029
Andrew Park: And because of Chat Gpt.

115
00:18:07.240 --> 00:18:15.500
Andrew Park: he found out because he's a very resourceful guy that he was able to do all that software development himself, even though he's a civil engineer.

116
00:18:16.010 --> 00:18:26.369
Andrew Park: And he was able to also in the end get a job done in 3 and a half days. That used to take 2, 2 and a half to 3 months.

117
00:18:26.580 --> 00:18:28.929
Andrew Park: And so it's an amazing story

118
00:18:29.220 --> 00:18:38.930
Andrew Park: of software engineering jobs being absorbed by someone that's really doesn't have a strong technical background. But it's very resourceful.

119
00:18:40.345 --> 00:18:43.020
Andrew Park: Now in that video

120
00:18:43.130 --> 00:19:04.010
Andrew Park: he shares something where he says he tells all of his friends, do not go into computer science. There's not going to be any of those jobs in the future. And that's not really going to be true. Okay, it's not as bad as what he thinks. He has that perspective, because the kinds of software that his team needed to write were actually pretty simple stuff.

121
00:19:04.735 --> 00:19:25.339
Andrew Park: what Gen. AI is gonna do is it's just gonna raise the bar of the level that each software developer can contribute at. And so it's it's basically if you get power tools, if you have power saws and other kinds of power tools. Well, you can produce much more than you could with hand tools. That's what the situation is.

122
00:19:25.590 --> 00:19:50.239
Andrew Park: So I would recommend looking into that story. But not to be too freaked out by his projections of the software development field going away. That's not really what's gonna happen. But what is happening is that companies that are adopting Gen. AI for software development, they are able to get a lot more done with a lot less people.

123
00:19:50.480 --> 00:20:17.610
Andrew Park: And so at Eden, soft labs, we've actually done a lot of analysis of open source code bases. And we've done developer productivity analysis of 420 companies covering 17 billion lines of code and 820,000 full-time engineers. And this is what we see that there's significant increase in developer productivity amongst companies that have heavily adopted this.

124
00:20:18.195 --> 00:20:25.409
Andrew Park: And these companies are examples, Microsoft, Google, Amazon, Meta. There's others, too. But

125
00:20:25.610 --> 00:20:30.249
Andrew Park: what's happened is that there has been a decrease in headcount via layoffs.

126
00:20:30.490 --> 00:20:41.449
Andrew Park: but also at the same time rising profits due to greater financial efficiencies. So this is a chart here that I'm showing of these 4 companies

127
00:20:42.131 --> 00:20:58.560
Andrew Park: the headcount decreasing, but at the same time their profits increasing. So the sum of the profits of these companies that are, you know, basically providers of Gen. AI, and then therefore definitely making their developers use it.

128
00:20:58.830 --> 00:21:01.730
Andrew Park: Their profits are at all time highs.

129
00:21:02.270 --> 00:21:07.929
Andrew Park: And so there is a trend doing more with less.

130
00:21:08.190 --> 00:21:24.189
Andrew Park: And I don't think that these statistics are things that are really widely known. I want you all to know about it, so that you know that it is definitely true that there is a lot of impact that Gen. AI is happening.

131
00:21:25.090 --> 00:21:34.620
Andrew Park: Alright. So here's the impact on employability. This is what you need to know as of today.

132
00:21:34.750 --> 00:21:39.810
Andrew Park: Gen. AI is very effective for performing many tasks.

133
00:21:40.030 --> 00:21:43.609
Andrew Park: but is not very effective for displacing entire jobs.

134
00:21:44.780 --> 00:21:52.129
Andrew Park: Gen. AI productivity boosts are driving these tech layoffs due to the need for less humans to get the work done.

135
00:21:52.860 --> 00:22:17.229
Andrew Park: But we're also seeing that probably more than 80% of the developers out there that we've analyzed are really lagging in Gen. AI adoption and proficiency. It's like really a small fraction of companies. That is obvious that they've really pushed this onto their developers. And they're seeing really big gains. And we are, too. We are, too. On on my team.

136
00:22:17.580 --> 00:22:28.309
Andrew Park: But mastering Gen. AI is key to current and future employability. So this is something that I hope at the end of this.

137
00:22:28.450 --> 00:22:42.630
Andrew Park: that you all are motivated as performers to really check out how to get good at this, so that you can be very viable as a employee in the future.

138
00:22:42.930 --> 00:22:47.300
Andrew Park: Okay, so here's the good news about Gen. AI.

139
00:22:48.167 --> 00:22:52.040
Andrew Park: I. This is how I describe it to my clients.

140
00:22:52.600 --> 00:22:56.140
Andrew Park: I'll say that Gen. AI is the 1st time

141
00:22:56.260 --> 00:23:03.030
Andrew Park: that we have had the ability to scale human talent. I mean, I'm talking about just about anybody.

142
00:23:03.480 --> 00:23:07.880
Andrew Park: and you can scale people horizontally.

143
00:23:08.210 --> 00:23:14.359
Andrew Park: meaning that it's almost like that. There's multiple copies of me. We can get more work done

144
00:23:15.464 --> 00:23:18.509
Andrew Park: so there's amplified productivity. That's what I showed you before.

145
00:23:18.620 --> 00:23:21.589
Andrew Park: But then there's also vertical amplification

146
00:23:21.840 --> 00:23:40.210
Andrew Park: where you can use Gen. AI to bridge the novice to Expert Gap. It's possible more than ever for novices to produce work. That's way above what you think that they should be able to produce with the assistance of AI. So this is really really exciting.

147
00:23:41.950 --> 00:23:49.809
Andrew Park: Also, because of Gen. AI, it's possible to learn new skills more easily than ever.

148
00:23:49.980 --> 00:23:56.750
Andrew Park: because encoded in all these models from these big tech companies is basically the world's recorded knowledge.

149
00:23:56.950 --> 00:24:04.549
Andrew Park: and and pretty much by now they have sucked in about everything that they're legally allowed to do, I mean, so

150
00:24:04.640 --> 00:24:22.860
Andrew Park: so there! And there isn't a huge difference between, let's say, Anthropic's best or chat Gpt's best models. They're all very similar, because they've all been very, very good at absorbing the world's available knowledge.

151
00:24:23.187 --> 00:24:41.500
Andrew Park: And now, because they're kind of on an even playing field, you're gonna see lots of deals made for data. And so recently, Reddit got this huge deal from one of those companies where they had to pay a lot of money. I think it was Google to be able to suck in their data

152
00:24:41.500 --> 00:24:56.899
Andrew Park: to make their model better. So you're going to see a lot of this kind of data deals happening. And then that will give some of these models advantages over others. And it's important to know those kinds of things in the future. So that you're using the right model for your task.

153
00:24:58.170 --> 00:25:04.680
Andrew Park: Okay, so Gen. AI is going to have an impact on every career.

154
00:25:04.800 --> 00:25:15.950
Andrew Park: And on this chart here, I'm showing 5 particular professions or roles. There's software developers here, designers.

155
00:25:16.310 --> 00:25:21.069
Andrew Park: scrum masters, product owners and product managers.

156
00:25:21.608 --> 00:25:30.329
Andrew Park: We have materials for all the other roles, too. But I'm limiting this to certain roles that I know are attending this talk.

157
00:25:30.820 --> 00:25:31.810
Andrew Park: Now

158
00:25:33.090 --> 00:25:54.740
Andrew Park: the bigger picture is that everything again, that you see here in red and yellow is being highly impacted by Gen. AI. So those tasks you can. If you're in this role, you can use Gen. AI or Gen. AI tools to be able to boost your productivity in all of these roles.

159
00:25:55.190 --> 00:25:59.839
Andrew Park: And that is what is going to continue to happen across the world.

160
00:26:00.130 --> 00:26:12.260
Andrew Park: and that will be very exciting and empowering on one hand. But it's also gonna drive layoffs. So we're going to see, continued layoffs of many, many roles in the future.

161
00:26:12.390 --> 00:26:16.579
Andrew Park: Okay, but that doesn't necessarily mean everyone's going to be unemployed.

162
00:26:16.800 --> 00:26:32.290
Andrew Park: There's going to be a lot of pressure to become cross functional. So this is not all bad news. Okay? So there's glass F, half empty glass, half full type of perspectives. I want you to see this as glass half full also.

163
00:26:32.630 --> 00:26:39.119
Andrew Park: So the glass half full is that if you are a scrum master or you're a product owner, let's say

164
00:26:39.460 --> 00:26:53.660
Andrew Park: the fact that you see a lot of red and yellow on these other roles here means that the barrier to entry of learning how to do those things and contributing at work on those functions has gone down massively.

165
00:26:55.620 --> 00:27:04.520
Andrew Park: So this is how Gen. AI not only forces a move to being cross, functional, greatly empowers it.

166
00:27:05.290 --> 00:27:11.850
Andrew Park: And so this is the kind of thing that I've been pushing on my people for a long time because I could. I could see.

167
00:27:11.970 --> 00:27:31.339
Andrew Park: Once I started playing with Chat Gpt. 3.5 a couple of months after it came out I was just like, Oh, my gosh! I can't believe this, and and it become very obvious to me that the barrier to entry, to be more and more cross, functional had massively gone down.

168
00:27:31.440 --> 00:27:41.340
Andrew Park: And and so this is a huge, huge advantage for us as humans to be able to broaden and deepen ourselves.

169
00:27:41.450 --> 00:27:51.010
Andrew Park: And it's a huge advantage for people who are leading businesses also to be able to do a lot more with a lot less. So

170
00:27:51.190 --> 00:27:57.200
Andrew Park: this is something where everyone can win, whether it's the business or the individual performer.

171
00:27:57.750 --> 00:28:06.460
Andrew Park: Now, a key insight. I'm going to show you here. Why are certain things green and still in the human camp, and a lot of other things are not

172
00:28:06.730 --> 00:28:10.140
Andrew Park: well, it's actually pretty simple. Here's the key insight.

173
00:28:10.300 --> 00:28:16.760
Andrew Park: Since Gen. AI consumes it's text audio images and video.

174
00:28:16.980 --> 00:28:21.900
Andrew Park: it's impacting most computer screen related tasks done by humans. Okay.

175
00:28:22.280 --> 00:28:27.450
Andrew Park: so so you could actually probably go to just about any profession.

176
00:28:27.810 --> 00:28:35.080
Andrew Park: and you could probably make a mind map and then predict which those things are going to be heavily impacted by Gen. AI,

177
00:28:35.660 --> 00:28:43.480
Andrew Park: and so there is massive changes happening in the world. It's happening very, very quickly

178
00:28:43.975 --> 00:28:51.439
Andrew Park: but don't be afraid. Now's the time to really get on the ball and take advantage of this and evolve yourself.

179
00:28:52.610 --> 00:28:59.029
Andrew Park: Okay, now, let's get into Gen. AI strengths and limitations.

180
00:29:01.700 --> 00:29:02.580
Andrew Park: All right.

181
00:29:03.310 --> 00:29:06.509
Andrew Park: So as of January 2025

182
00:29:07.180 --> 00:29:11.549
Andrew Park: can AI models excel at handling specific tasks.

183
00:29:12.380 --> 00:29:17.059
Andrew Park: but they fall short when it comes to fully automating most human jobs

184
00:29:17.650 --> 00:29:33.819
Andrew Park: so they can perform well for focused activities, like generating summaries or reviewing documents, writing code snippets. But they actually require a lot of human iteration and validation for complex workflows.

185
00:29:34.050 --> 00:29:42.069
Andrew Park: So I advise my clients to think of Gen. AI. Models as an unlimited pool of specialized workers.

186
00:29:42.260 --> 00:29:53.320
Andrew Park: where each one is capable of completing narrow tasks, but requiring clear, explicit instructions and careful validation to ensure the desired results.

187
00:29:53.810 --> 00:29:58.455
Andrew Park: So I list here 5 current limitations of Gen. AI.

188
00:30:01.820 --> 00:30:05.369
Andrew Park: The 1st one is limited context windows.

189
00:30:05.830 --> 00:30:09.880
Andrew Park: So, for example, if you're a product owner

190
00:30:10.220 --> 00:30:16.229
Andrew Park: and us. Gen. AI. To summarize a large product requirements document, the model

191
00:30:16.410 --> 00:30:21.670
Andrew Park: may only process part of that content leading to missing key sections

192
00:30:22.241 --> 00:30:28.420
Andrew Park: like critical acceptance, criteria or regulatory requirements, you might be be disappointed in the results.

193
00:30:29.250 --> 00:30:33.289
Andrew Park: The second limitation is lack of real world understanding.

194
00:30:33.720 --> 00:30:48.920
Andrew Park: So again, if you're a product owner, you could ask the model to generate a user story for a highly regulated financial product. But the output may overlook critical compliance requirements, since the model actually lacks true domain knowledge.

195
00:30:50.410 --> 00:30:54.379
Andrew Park: The 3rd limitation is task specialization.

196
00:30:55.110 --> 00:31:02.530
Andrew Park: So a designer might use Gen. AI to generate creative ideas for a homepage layout.

197
00:31:02.840 --> 00:31:08.180
Andrew Park: but would struggle to get cohesive design recommendations for an entire design system.

198
00:31:09.250 --> 00:31:14.570
Andrew Park: And the 4th limitation is decision complexity.

199
00:31:14.900 --> 00:31:21.890
Andrew Park: So, for example, if you're a scrum master, you could use Gen. AI to suggest how to resolve a team conflict.

200
00:31:22.290 --> 00:31:29.819
Andrew Park: But the model would lack emotional intelligence, and it would lack context needed to make sound judgment calls in human dynamics.

201
00:31:30.060 --> 00:31:30.800
Andrew Park: Okay?

202
00:31:30.990 --> 00:31:48.000
Andrew Park: And the 5th limitation is dependence on human validation and iteration. And I'm going to talk a lot about this in the rest of the talk. So we run into this all the time when working with Chat Gpt or other Gen. AI models.

203
00:31:48.140 --> 00:32:05.039
Andrew Park: So, for example, if you're a developer, you might generate unit tests using chat Gpt, but those tests will often require human validation and review to make sure they cover edge cases and fully align with your project's quality standards

204
00:32:05.310 --> 00:32:13.100
Andrew Park: and without manual refinement the results will definitely have all kinds of gaps.

205
00:32:14.325 --> 00:32:17.729
Andrew Park: So while Gen. AI can be a powerful tool.

206
00:32:18.060 --> 00:32:27.359
Andrew Park: its limitations highlight, the need for human oversight, clear instructions and validation to ensure that you get the results you want.

207
00:32:28.150 --> 00:32:30.449
Andrew Park: Okay, now, let's talk about this table here.

208
00:32:31.080 --> 00:32:34.680
Andrew Park: This table shows a range of Gen. AI models

209
00:32:35.100 --> 00:32:51.879
Andrew Park: and smaller models like Gpt. 3.5 and Gpt. 4, standard. They work best for simple tasks with limited context windows. The 1st line here or 1st column. Here is a context window. This is in number of tokens.

210
00:32:52.120 --> 00:33:07.749
Andrew Park: And so Gpt. 3.5 had a context window of 4,000 tokens that translates to about 5,000 words. If you're processing text, okay, if you're doing audio or video, it's it's different. But just to keep it simple. I'm just going to talk about text.

211
00:33:08.020 --> 00:33:12.140
Andrew Park: all right. And then Gpt. 4. Standard doubled that.

212
00:33:12.260 --> 00:33:23.719
Andrew Park: But now you go to. Today's Gpt. 4, and that's 128,000. And so you can see that these have been greatly increasing over time.

213
00:33:24.110 --> 00:33:26.570
Andrew Park: and this is a core spec

214
00:33:26.750 --> 00:33:48.749
Andrew Park: for the Gpt models or or the Gen. AI models. It's 1 of the reasons why I tell everybody it's worth paying for one of the good ones, you know. Don't just use a free one that has really limited context window. You're you're not going to be able to get good results out of that or competitive results.

215
00:33:49.330 --> 00:34:10.790
Andrew Park: You can see at the top end here for context window is Gemini, 1.5 pro where it actually has a 2 million tokens. So it's like 2.6 million words. It's enormous compared to the rest, and I'm going to talk later about certain kinds of tasks where you need to use something like Gemini, and you can't. You don't have any other choice.

216
00:34:12.540 --> 00:34:16.690
Andrew Park: The next column here is cost per mega token.

217
00:34:17.449 --> 00:34:22.199
Andrew Park: And so you can see that these have very different costs.

218
00:34:23.080 --> 00:34:28.750
Andrew Park: And then the last column is task complexity. So

219
00:34:28.989 --> 00:34:33.159
Andrew Park: I'm hoping that this will really help you all be able to get

220
00:34:33.310 --> 00:34:49.309
Andrew Park: an idea of how to compare and contrast the use of these models, and if you understand this table really well, this will give you a lot of insights into why, some people are getting better results than others for certain tasks.

221
00:34:49.620 --> 00:34:52.750
Andrew Park: So all right.

222
00:34:57.120 --> 00:34:58.550
Andrew Park: Okay, so

223
00:35:00.150 --> 00:35:08.680
Andrew Park: mid tier models like, what we generally use with Chat Gpt, which is based. It's similar. Gpt for Turbo

224
00:35:08.970 --> 00:35:18.169
Andrew Park: and Claude. 3 sonnet is another example, a balance context, size and cost for giving moderate complexity tasks.

225
00:35:18.430 --> 00:35:32.540
Andrew Park: and then larger models like Claude Claude, 3 opus and gemini, 1.5 pro. They handle very large data sets if you have that and strategic tasks with high capacity. But higher costs.

226
00:35:33.130 --> 00:35:41.760
Andrew Park: So choosing the right model is about balancing the context size that you need the task complexity and the cost efficiency

227
00:35:42.080 --> 00:35:46.679
Andrew Park: and smaller models. Will fit

228
00:35:47.060 --> 00:35:56.010
Andrew Park: quick focus tasks, while larger models are really essential for you to use if you are dealing with complex data, heavy work.

229
00:35:56.650 --> 00:35:57.480
Andrew Park: Okay?

230
00:35:57.590 --> 00:36:02.869
Andrew Park: So now let's explore how these models apply to specific roles and product teams.

231
00:36:04.214 --> 00:36:06.819
Andrew Park: Alright so. But 1st of all.

232
00:36:07.390 --> 00:36:09.809
Andrew Park: I want to give a warning here

233
00:36:10.150 --> 00:36:14.620
Andrew Park: that, Jenny I output should never be blindly copy, pasted

234
00:36:14.820 --> 00:36:22.339
Andrew Park: without validation and iteration, the results can be misleading, generic, or even harmful.

235
00:36:22.410 --> 00:36:47.729
Andrew Park: So in this graphic here, I'm saying that this is the keyboard associated with the Gen. AI horror stories that you've read. Okay, there's a lot of these stories circulating on social media that really kind of debunk people's interests and in using it like, Oh, yeah, it sucks for this sucks for that. Did you hear about this? Did you hear about that horrible thing that happened, and I really really want to encourage you.

236
00:36:47.850 --> 00:36:56.009
Andrew Park: Do not let that influence you into thinking that you should not be aggressively using AI all day, every day.

237
00:36:56.508 --> 00:37:10.529
Andrew Park: It really bothers me when I see those stories, because I know the effect on people is like, Oh, this is something I just have to worry about later. No, you got to worry about it now. It has great, great, great utility now.

238
00:37:10.710 --> 00:37:24.449
Andrew Park: and so I'm hoping that this talk will really motivate everybody to be able to understand why those stories are true, while at the same time it should not affect your enthusiasm to go after it right now.

239
00:37:25.040 --> 00:37:26.100
Andrew Park: So.

240
00:37:29.340 --> 00:37:34.079
Andrew Park: using the AI right now, senior level professionals with their expertise.

241
00:37:34.340 --> 00:37:39.940
Andrew Park: they know how to refine AI generated content for quality and originality.

242
00:37:40.560 --> 00:37:54.199
Andrew Park: And this is why many Silicon Valley companies have shifted away from junior hiring. They're retaining the senior people that have the skill to guide and correct AI effectively.

243
00:37:54.380 --> 00:37:57.540
Andrew Park: So it's more essential now than ever

244
00:37:57.750 --> 00:38:08.019
Andrew Park: for workers to amass both both problem domain knowledge and solution, domain knowledge. And that's something that I'm going to be emphasizing later in this talk, too.

245
00:38:09.150 --> 00:38:10.120
Andrew Park: Okay.

246
00:38:17.565 --> 00:38:19.914
Andrew Park: one of the things that I think

247
00:38:20.670 --> 00:38:22.880
Andrew Park: is likely to happen in the future

248
00:38:23.050 --> 00:38:26.479
Andrew Park: is that we're going to see that

249
00:38:27.400 --> 00:38:54.960
Andrew Park: almost everything that we buy, whether it's a hardware device or software application in the future. We're actually going to probably have subscriptions to all kinds of AI agents or or things powered by AI agents. And it's gonna need to be able to use computing resources. Gen. AI computing resources. I could see in the future us being able to

250
00:38:55.060 --> 00:39:14.849
Andrew Park: have to buy Gen. AI computing resources as a utility just like we do. For electricity. So just like how you get billed by the kilowatt for electricity, we might get billed by the Mega token by utility providers for Gen. AI.

251
00:39:18.060 --> 00:39:24.180
Andrew Park: Now, when choosing a Gen. AI model for developers. Here's some tips.

252
00:39:26.070 --> 00:39:29.819
Andrew Park: Gpt, 4 turbo or current chat. Gpt

253
00:39:29.920 --> 00:39:58.950
Andrew Park: is a pretty reasonable choice for code generation and mid-level code reviews. But it actually is pretty limited for unit test generation. It will miss a lot of education cases and broader test coverage. That's why, later on. I'm going to recommend a different tool that actually uses Gen. AI. But it'll do a lot better job than than what you'll get out of Gpt for

254
00:39:59.486 --> 00:40:07.770
Andrew Park: Claude. Opus cloud 3 opus. It offers stronger reasoning, making it better suited for generating more complete unit tests.

255
00:40:08.287 --> 00:40:17.209
Andrew Park: While Gemini, 1.5 pro is your only choice when working with large code bases where comprehensive test coverage is is required.

256
00:40:17.490 --> 00:40:24.850
Andrew Park: and I'm sure that the best tools that are doing unit test coverage are are using these one of these.

257
00:40:25.510 --> 00:40:26.420
Andrew Park: So

258
00:40:27.131 --> 00:40:42.009
Andrew Park: for complex code reviews and system architecture analysis. Claude, 3 opus provides deeper reasoning, whereas lesser models will be likely to miss architectural patterns and cross module dependencies.

259
00:40:42.120 --> 00:41:01.190
Andrew Park: So the moral of the story is that if you need, you need to select the model that best fits the task. Complexity and scale. If you want good results, or, alternatively, if you don't want to have to do everything manually yourself. You can adopt an AI tool from vendor, and then it's that's hitting the easy button

260
00:41:04.640 --> 00:41:06.510
Andrew Park: which I think most people probably do

261
00:41:06.940 --> 00:41:09.600
Andrew Park: alright. Now let's talk about for designers

262
00:41:10.859 --> 00:41:32.820
Andrew Park: for designers. Plot 3 sonnet. It's a good choice for straightforward tasks, like giving feedback on design drafts or suggesting creative ideas, or helping with ux guidelines, and it works well when you need quick suggestions for smaller projects. But it may miss details when reviewing more complex design systems.

263
00:41:34.980 --> 00:41:44.860
Andrew Park: if you need deeper analysis, flaw 3 opus is better for reviewing entire design systems and working across multiple products.

264
00:41:45.120 --> 00:41:51.350
Andrew Park: For example, if a company is redesigning both its mobile app and a web dashboard.

265
00:41:51.480 --> 00:42:12.849
Andrew Park: A design system review would need to ensure consistent use of colors, button styles and spacing across both both platforms, and you don't want to use claw 3 sonnet for that. You want to try to use Claude 3 opus, because it can help you identify mismatch patterns or inconsistencies in how design elements are applied across those interfaces.

266
00:42:13.630 --> 00:42:20.170
Andrew Park: Now for the most complex design challenges. I'd recommend trying out Gemini 1.5 pro

267
00:42:20.671 --> 00:42:38.670
Andrew Park: it would be the strongest option. It can handle system, wide design audits and review large design libraries for consistency. That makes it ideal when working on multiple products, such as a suite of productivity tools where every application must share the same design language.

268
00:42:39.990 --> 00:42:40.850
Andrew Park: So

269
00:42:41.230 --> 00:42:59.739
Andrew Park: the right model depends on the scope of your project. Again, smaller models are fine and cheap and great. If you want quick creative input, while the more advanced models are kind of necessary if you want to use them for design consistency analysis across large and complex products.

270
00:43:03.270 --> 00:43:05.689
Andrew Park: Okay, now, let's talk about product owners.

271
00:43:06.810 --> 00:43:35.829
Andrew Park: So product owners. Gpt for Turbo or today's Gpt is a reasonable choice for writing detailed feature breakdowns and acceptance criteria. You can use it to help draft user stories. For, let's say, a login page redesign could outline requirements such as multi factor, authentication, error, messaging and password, reset functionality.

272
00:43:36.622 --> 00:43:42.510
Andrew Park: But it will definitely require review to ensure that nothing critical is overlooked.

273
00:43:42.780 --> 00:44:00.530
Andrew Park: so you can't again just copy paste what it gives you. You need to use your human expertise to validate and iterate to get the good results. It's going to Miss Edge cases. But on the balance of things it's going to really help your productivity.

274
00:44:01.528 --> 00:44:05.440
Andrew Park: Now, Claude, 3 opus offers deeper reasoning.

275
00:44:05.550 --> 00:44:12.170
Andrew Park: and so it makes it better suited for strategic planning across product portfolios and Kpi reporting.

276
00:44:12.620 --> 00:44:20.690
Andrew Park: And so, for instance, as a product owner, if you're working on multiple product enhancements. Within the same platform

277
00:44:20.850 --> 00:44:26.199
Andrew Park: flawed 3 opus can assist in prioritizing features by business impact.

278
00:44:26.330 --> 00:44:34.950
Andrew Park: and it can identify overlapping customer needs or help you align kpis, like feature, adoption, and churn reduction across the roadmap.

279
00:44:36.230 --> 00:44:38.830
Andrew Park: Now, Gemini, 1.5 pro

280
00:44:39.545 --> 00:44:49.180
Andrew Park: is a good choice for doing strategic alignment and long term planning and market analysis. So, for example.

281
00:44:49.690 --> 00:45:08.650
Andrew Park: if as a product owner, you want to pair a quarterly planning session, it can help forecast how upcoming features might affect customer retention, assistant aligning priorities across multiple product teams and even surface insights from historical data to inform market positioning strategies.

282
00:45:09.020 --> 00:45:28.669
Andrew Park: So again, choosing the right model depends on the complexity of your planning task. Gpt-four Turbo works well for smaller tactical tasks like story writing while clothy opus and gemini, 1.5 pro offer stronger support for broader strategic planning and multi-product coordination.

283
00:45:34.190 --> 00:45:36.790
Andrew Park: Okay, now for scrum masters

284
00:45:38.416 --> 00:45:45.789
Andrew Park: Gpt for Turbo. It's a reasonable choice for your everyday tasks like sprint, reporting

285
00:45:45.940 --> 00:45:49.739
Andrew Park: retrospective summaries and drafting process documents.

286
00:45:50.140 --> 00:45:54.809
Andrew Park: So as an example, you can use it to help summarize a sprint review

287
00:45:54.960 --> 00:46:01.380
Andrew Park: by listing completed work open tasks and blockers in a clear format for the team and stakeholders.

288
00:46:01.910 --> 00:46:15.270
Andrew Park: But it will require adjustments to ensure that the summary captures, the team discussions and context accurately. Again, don't copy paste, guide it, validate it, iterate with it.

289
00:46:16.250 --> 00:46:18.050
Andrew Park: Clothury. Opus

290
00:46:19.024 --> 00:46:34.359
Andrew Park: would be a better suited for your more complex tasks. Like if you're doing safe and you want to do safe pi planning or manage cross team dependencies looking to claw 3 Opus. So

291
00:46:34.610 --> 00:46:50.390
Andrew Park: so imagine you're facilitating a pi planning session with multiple teams working on related features claw. 3 opus can help map out how one team's deliverable impacts, another such as ensuring that an Api is ready before a front end, features built.

292
00:46:50.640 --> 00:46:57.749
Andrew Park: but it still needs the scrum master's judgment to finalize the timeline and assess the team capacity

293
00:46:59.585 --> 00:47:07.669
Andrew Park: Gemini. 1.5 pro could help you in managing large scale depend dependencies and reporting across multiple teams.

294
00:47:07.900 --> 00:47:08.880
Andrew Park: For example.

295
00:47:08.990 --> 00:47:29.109
Andrew Park: if you're coordinating multiple scrum teams in a large product release Gemini, 1.5 Pro. Could help generate a visual overview showing which teams are working on what it could. Highlight blockers across the program and assist with preparing executive level reports on overall progress and risk areas.

296
00:47:29.230 --> 00:47:32.660
Andrew Park: So the right model again depends on the task.

297
00:47:32.770 --> 00:47:48.150
Andrew Park: Gpt, 4 Turbo works well for your everyday reporting stuff, while the cloud, 3 opus and Gemini, 1.5 pro. Would be better for your strategic planning and managing complex multi-team coordination.

298
00:47:50.640 --> 00:47:51.670
Andrew Park: Okay?

299
00:47:52.760 --> 00:48:17.059
Andrew Park: So that last part was all about assuming that you don't have vendor AI tools. And you're just doing everything yourself from the Gen. AI prompt screen. You can do a lot if you understand how to do prompt engineering. Well, and I definitely recommend everybody to take a prompt engineering course, Google has a really nice one, and it's not expensive.

300
00:48:18.610 --> 00:48:48.050
Andrew Park: you can get a lot of utility out of Gen. AI, even if your company leadership does not invest in all these fancy tools that I'm gonna talk about here. Okay. So whether you have these fancy tools or not is really often not up to you. It's up to your your upper management. But even if they don't do that, you should get very good at prompt engineering and use Gen. AI in the kind of ways and tasks that I covered before.

301
00:48:48.180 --> 00:48:49.415
Andrew Park: Okay,

302
00:48:51.620 --> 00:49:02.850
Andrew Park: alright. So now I'm gonna talk about tools. I'm gonna talk about a variety of tools, and for everything that I have here.

303
00:49:03.580 --> 00:49:11.370
Andrew Park: I'm gonna talk about things that I'm familiar with, things that I've or things that I've heard, you know very good things about

304
00:49:11.942 --> 00:49:32.709
Andrew Park: it doesn't mean that it's the only choice. Okay? So just because I have chat gpt on here doesn't mean that you can't use anthropic. No, it worked totally fine. You could, you, could you? There's alternatives to everything, right? So you could go to chat, Gpt and say, Oh, what are the alternatives to this tool? And then, oh, what are the pros and cons? What does it do? What does it not do compared to the rest.

305
00:49:32.710 --> 00:49:43.820
Andrew Park: You know that that's something you can very easily do. I think it's smart for you to do that in order to know what's available on the market because there are a ton of things available on the market.

306
00:49:44.561 --> 00:49:59.010
Andrew Park: Right now. There has been a ton of venture capital going into lots of little startups making tools that are built on top of Gen. AI. And all of these are examples of that.

307
00:49:59.130 --> 00:50:14.999
Andrew Park: But there's just a plethora. So that's something that's really nice. So so I'm talking to talk about just certain ones just to keep the description easy. But there's a lot of choice out there.

308
00:50:15.290 --> 00:50:33.530
Andrew Park: Okay, so let's look at this. Mind map, and highlighting in the mind map certain things. So in this case, code, implementation and documentation we're dealing with software developers. Here again, this is a mind map of all the different kinds of tasks that the software developer

309
00:50:34.144 --> 00:50:57.579
Andrew Park: deals with. And then what I'm showing in here is that I'm highlighting certain tools that help with certain tasks. So I'm talking. Gonna talk about Chat Gpt github, copilot and cursor, as they relate to accelerating code implementation and documentation. Okay? So I just wanted to describe that because I'm gonna be going through a ton of these slides.

310
00:50:58.170 --> 00:50:58.960
Andrew Park: All right.

311
00:50:59.250 --> 00:51:06.540
Andrew Park: So Gen. AI plays a significant role in both code implementation and documentation

312
00:51:07.234 --> 00:51:15.150
Andrew Park: and our developers make extensive use of github copilot, and it excels in inline code suggestions.

313
00:51:15.600 --> 00:51:25.960
Andrew Park: completing functions, generating basic documentation directly within the ide based on prompts and the surrounding code context.

314
00:51:26.160 --> 00:51:37.710
Andrew Park: So it's a very, very popular tool. It's very empowering for every developer. business leaders, you you've got to adopt something like this. It'd be crazy not to.

315
00:51:37.940 --> 00:51:48.210
Andrew Park: And it significantly accelerates coding tasks by providing relevant code completions and reducing boilerplate work across various languages and frameworks.

316
00:51:48.754 --> 00:51:50.759
Andrew Park: I also want to mention cursor

317
00:51:50.880 --> 00:52:14.140
Andrew Park: cursor is something newer, and it's gaining traction. As an alternative to copilot, it offers some advantages. It's not like world's different or worlds better. It's it's got some advantages. It offers faster code based searches and more accurate multi file editing, making it particularly effective for navigating large projects and refactoring code bases.

318
00:52:14.150 --> 00:52:20.459
Andrew Park: And one thing I like about first, st or just qualitatively, is that

319
00:52:21.780 --> 00:52:33.309
Andrew Park: it? It's kind of has its own id, you know. It has its own native id rather than copilot. Which kind of integrates with all your existing ids

320
00:52:33.717 --> 00:52:56.929
Andrew Park: and my team hasn't switched to cursor at this time. I don't know if we will. We're gonna try it out. I'll make the decision later. I'll probably share on my linkedin and social media. Whether we kept copilot or went to cursor. I know that source graph wants me to try out their stuff, too. We might try that out, too, but there's lots of choices, and there's a lot of good choices.

321
00:52:58.247 --> 00:53:01.169
Andrew Park: Now, I've also highlighted chatgpt.

322
00:53:01.280 --> 00:53:06.399
Andrew Park: So our developers also frequently use chat Gpt, even though we have copilot

323
00:53:06.590 --> 00:53:20.629
Andrew Park: and chatgpt is highly effective for explaining complex code concepts, generating code from scratch based off of natural language prompts and providing detailed debugging assistance. Now.

324
00:53:22.100 --> 00:53:40.939
Andrew Park: while Chatgpt offers broader technical insights and reasoning, it lacks tight ide integration and real time code context, awareness that copilot and cursor could give you. And so it's less suited. It's not really well suited for direct in editor coding type of support.

325
00:53:42.260 --> 00:53:43.250
Andrew Park: But

326
00:53:43.270 --> 00:54:12.109
Andrew Park: sadly, none of these tools excel at generating documentation that effectively conveys the coders intent. That's the bad news. Okay, with AI enabling developers to produce code far more rapidly. The volume of code generated can create a significantly greater maintenance burden for human developers and AI tools can't fully compensate for this increased complexity because

327
00:54:12.400 --> 00:54:35.829
Andrew Park: they actually lack a true understanding of long term design principles. They lack of understanding of architectural consistency, and they also have no idea what's the deeper intent behind these coding decisions? Right? And so the humans are still going to be very involved in maintenance, unfortunately, and there's going to be a lot more maintenance to be done. So for

328
00:54:36.140 --> 00:55:03.570
Andrew Park: so while AI can assist with task level suggestions and localized refactoring. It struggles with enforcing maintainability patterns across entire systems or evolving code bases over time, and this makes it more critical than ever for developers to learn techniques for writing highly maintainable code, because human expertise remains essential for long-term clarity, sustainability and reducing technical debt.

329
00:55:04.660 --> 00:55:05.359
Andrew Park: So

330
00:55:06.040 --> 00:55:26.300
Andrew Park: we do a lot to train our developers in coding technique because of this. And we're we've always done this. But but we have not stopped. And the rise of Gen. AI has not taken away the need to do that. It's actually kind of almost increased it, because there's more more code to maintain.

331
00:55:26.500 --> 00:55:31.540
Andrew Park: Now, one of the things that we have available

332
00:55:32.640 --> 00:55:39.529
Andrew Park: just recently, here is the coding technique training that we've developed in house.

333
00:55:40.225 --> 00:55:59.809
Andrew Park: That's more extensive than the resources and publicly available out there. This is something I'm going to make available to all agile alliance members. If you want to access this where there's tons of coding technique tips

334
00:55:59.810 --> 00:56:17.370
Andrew Park: that are very, very practical things that your code mentors or code reviewers could use to train people in writing really maintainable code, you can email support@edensoftlabs.com to request access. And you can get access to this

335
00:56:20.380 --> 00:56:21.269
Andrew Park: all right.

336
00:56:21.790 --> 00:56:24.240
Andrew Park: The next thing is code review.

337
00:56:24.660 --> 00:56:30.240
Andrew Park: So for code review this tool, Codicey AI

338
00:56:30.670 --> 00:56:43.940
Andrew Park: is really useful. It can streamline code reviews by catching various bugs, vulnerabilities and code smells A nice feature is that it provides feedback directly in the pull requests

339
00:56:44.575 --> 00:56:50.529
Andrew Park: but one current weakness is that it has limited flexibility and customizing rule sets

340
00:56:51.157 --> 00:56:56.450
Andrew Park: but overall, it's a really solid option that is worth checking out.

341
00:56:56.960 --> 00:56:58.060
Andrew Park: And

342
00:56:58.480 --> 00:57:09.200
Andrew Park: one of one of the things that I will also say here is that although this is great for productivity, and the kind of

343
00:57:10.178 --> 00:57:16.159
Andrew Park: things that it does might be sufficient for some people's needs

344
00:57:16.240 --> 00:57:35.140
Andrew Park: if you are working on high consequence software. What I mean by that is software where, let's say, human life is in the balance or any kind of software where a bug could be horrible like cyber security or medical stuff. You know, aerospace things like this.

345
00:57:35.200 --> 00:58:04.579
Andrew Park: You should never just rely on these automated tools for code review. You need to make sure that you're cultivating really good human talent for being able to do reviews to make sure that horrible things don't slip into production the kinds of things that happened. You know, in recent history, with Crowdstrike is a real cautionary tale.

346
00:58:04.650 --> 00:58:08.539
Andrew Park: and those kinds of things are really

347
00:58:08.670 --> 00:58:22.430
Andrew Park: just you're going to hear more of it. More of these kinds of things are going to happen in high consequence applications. So it's really really important to to realize that again, don't use AI as copy paste

348
00:58:22.834 --> 00:58:41.050
Andrew Park: you need to cultivate human talent. That's very, very good at testing or very, very good at creating solid code. If you're doing high consequence software. Now, if you're doing stuff that's not like that, I don't know if you're doing restaurant recommendation. Yeah, maybe this will be good, though.

349
00:58:41.900 --> 00:58:47.150
Andrew Park: Okay, the next area is quality assurance.

350
00:58:47.570 --> 00:59:00.499
Andrew Park: So for quality assurance quota specializes in generating unit tests by using AI by analyzing code structure and suggesting test cases that cover various execution paths.

351
00:59:00.650 --> 00:59:09.159
Andrew Park: And it helps automate basic test coverage. Though the generated tests often require manual review to ensure that they're complete

352
00:59:09.430 --> 00:59:10.175
Andrew Park: and

353
00:59:11.050 --> 00:59:35.950
Andrew Park: it's much better than github, copilot or chatgpt and generating unit tests. Now for Gui testing a tool called test rigor uses Gen. AI to create end to end tests. And it's pretty cool. It observes your application behavior and user interactions. And then it generates test scripts based on those observations and allows users to define test scenarios in plain English.

354
00:59:35.950 --> 00:59:52.119
Andrew Park: So it makes it accessible to non-technical testers, and while it simplifies test creation and produces maintainable scripts, it may struggle with complex or highly dynamic uis, so expect to have human validation and iteration

355
00:59:55.840 --> 01:00:03.895
Andrew Park: for operations. Dynatrace has been a leader. There's lots of choices. Dynatrace is a competitor to data dog.

356
01:00:04.320 --> 01:00:32.530
Andrew Park: But they have been really trying to put in a lot of AI insights to help identify bottlenecks faster. But again, you know, for any of these observability tools expect that you're still going to need human validation for complex cases. But there, there's exciting things happening there in the observability space. And then for professional development and research and analysis.

357
01:00:32.984 --> 01:00:45.669
Andrew Park: Our team extensively uses chat Gpt, and feed the AI for professional development chat Gpt is our core research tool. You know, we're good at prompt engineering. But remember that

358
01:00:46.142 --> 01:00:51.759
Andrew Park: if you want to get use out of it, you you really should get good at prompt engineering.

359
01:00:52.177 --> 01:01:05.050
Andrew Park: Deeply. AI is something that complements this. It's it's a tool that we've been using for a long time, where you put in keywords of what your interests are, and then it forwards information to you.

360
01:01:05.399 --> 01:01:25.649
Andrew Park: It's useful, because you know who's got the time to scour the world for certain topics. So it helps keep our team informed on industry trends and best practices. You know, when I put on my engineering leadership hat, I wanna make sure that I'm aware of emerging technologies and tools. Because that drives innovation.

361
01:01:27.820 --> 01:01:33.460
Andrew Park: Okay, now, in regard to domain knowledge and soft skills.

362
01:01:34.164 --> 01:01:53.139
Andrew Park: Chat, Gpt's custom Gpts is a a feature that allows your domain experts to import knowledge from books, talks, or internal company documents to make it easier for developers to learn company, relevant domain knowledge through conversational guidance.

363
01:01:53.220 --> 01:02:13.310
Andrew Park: And this approach is really really similar to retrieval, augmented generation or rag where AI references external data to improve response accuracy. But custom Gpts rely on preloaded static data rather than retrieving live information.

364
01:02:13.570 --> 01:02:24.320
Andrew Park: And the advantage is that your developer teams will be able to receive consistent domain specific guidance simplifying complex concepts.

365
01:02:24.430 --> 01:02:31.249
Andrew Park: But since the data is static, your domain experts will need to periodically update it to keep the information current

366
01:02:31.980 --> 01:02:42.909
Andrew Park: so custom Gpts are a big step forward. They're a way to powerfully support professional development across teams. So if you're a leader of business,

367
01:02:43.710 --> 01:02:47.129
Andrew Park: this is something to really look into, and

368
01:02:47.160 --> 01:03:00.190
Andrew Park: this is something where Nvidia has really done a good job. Internally, I'm talking point internally. They've implemented a lot of things like these custom Gpts.

369
01:03:00.190 --> 01:03:28.630
Andrew Park: so that junior developers aren't bothering the senior developers. They're like for Nvidia internal knowledge. They can. They're trained to go 1st to these internal custom Gpts to get their questions answered. And it's really reduced the the load on Senior Level people having to mentor younger people. And it's something that definitely I pushed this internally here.

370
01:03:28.630 --> 01:03:54.970
Andrew Park: I don't like my junior developers bothering my senior developers. Because, like I said before, Gen. AI benefits senior people disproportionately to junior people. So senior people can be massively productive with Gen. AI and yeah, mentorship's important. But now that there can be an alternative to us, beginner and intermediate questions.

371
01:03:55.326 --> 01:03:59.609
Andrew Park: it just makes sense to offload that to to the machine.

372
01:04:01.150 --> 01:04:01.860
Andrew Park: Okay.

373
01:04:01.860 --> 01:04:02.500
Mohan Vemulapalli: Andrew, I.

374
01:04:02.500 --> 01:04:02.820
Andrew Park: No.

375
01:04:02.820 --> 01:04:07.840
Mohan Vemulapalli: Just want to let you know we're about an hour in, and we have a number of questions.

376
01:04:08.460 --> 01:04:09.190
Andrew Park: Okay.

377
01:04:09.490 --> 01:04:09.889
Mohan Vemulapalli: Thank you.

378
01:04:09.890 --> 01:04:35.110
Andrew Park: All right. Yeah. Okay, so let's talk about gen, AI impacts on designers the 1st one here, I want to talk about interface, prototyping, visual design, interaction design figma AI has introduced significant advancements and interface prototyping, visual design and interaction design. So now, with a simple text, prompt like.

379
01:04:35.330 --> 01:04:51.120
Andrew Park: create a mobile e-commerce, checkout screen with a product, summary payment options and a place order button. You could just give it to that, and it can generate something for you, and then you can refine it. After that. It's amazing.

380
01:04:51.616 --> 01:04:58.194
Andrew Park: So go on Youtube and go take a look at some of the Demos, and you'll see what I'm saying. It's

381
01:04:58.590 --> 01:05:04.880
Andrew Park: it's a it's a real step forward. It's something that I heavily, heavily promote to my clients.

382
01:05:05.110 --> 01:05:13.595
Andrew Park: So so the next thing all right, let me let me speed things up here.

383
01:05:14.810 --> 01:05:23.830
Andrew Park: the next thing is drafting user personas. So chat gpt, can significantly streamline the creation of user personas.

384
01:05:23.880 --> 01:05:45.000
Andrew Park: And you could summarize insights from user interview surveys and research data. And by using prompts like, create a user persona based on all these interview notes, including the pain points, the goals and behavior patterns. It can quickly generate detailed personas and reduce manual effort.

385
01:05:45.080 --> 01:05:56.289
Andrew Park: And so this can help teams identify audience segments faster while keeping a consistent format across personas. But of course, designers will still need to validate and iterate.

386
01:05:56.420 --> 01:06:05.370
Andrew Park: And and also, if you have a lot of this data, you might need to use one of those bigger models and models with a larger context windows.

387
01:06:06.850 --> 01:06:10.050
Andrew Park: okay, now, let's talk about visual design.

388
01:06:10.470 --> 01:06:26.870
Andrew Park: So for drafting website, content Jasper really stands out and it's got pre-built templates for headlines and product descriptions and call to action phrases and

389
01:06:26.990 --> 01:06:51.270
Andrew Park: it gives a structured approach to help designers and marketers quickly create engaging and conversion focused content while maintaining consistency across the project. Now, we've used Chat Gpt, to do this kind of stuff before, but it's not gonna give the same level of clarity and alignment that something that's really customized like Jasper will do.

390
01:06:52.910 --> 01:07:08.009
Andrew Park: Now for visual content. Creation runway is amazing. So go take a look at some of the videos that they have. And it can create highly realistic videos. From just mere text prompts. And

391
01:07:08.160 --> 01:07:11.360
Andrew Park: and there's some amazing demos out there.

392
01:07:11.520 --> 01:07:40.349
Andrew Park: Now, you definitely have to do lots of human validation and iteration if you care about something very specific. And so it's not as simple as like, you know, doing something in 5 seconds. But you can prototype something in 5 seconds very quickly. And so it's very, very useful for being able to do the you know, accelerated design and and brainstorming process. But you can get amazing results with it if you work with it.

393
01:07:40.490 --> 01:07:56.790
Andrew Park: and the same can be said with Dally I think most of you probably have played around with dally within chat gpt, and created all kinds of images. I do it all the time, both professionally and just for fun. And

394
01:07:57.470 --> 01:08:13.359
Andrew Park: so it it's really great at producing detailed images. But again, if you want something really, really specific, you'll find that it's really hard to get it to do exactly what you want. And so there are other tools emerging that are going to do probably a better job for certain domains.

395
01:08:13.980 --> 01:08:20.400
Andrew Park: But these tools are something that are, you know they're real game changers.

396
01:08:20.800 --> 01:08:45.359
Andrew Park: Now again, I want to just go back to. If you're a scrum master, if you're a product owner, if you're a project manager, or whatever it is. And I'm bring all these tools for developers and designers. Why do you care? Why, you might care? Is, did you ever want to do some of this work. Did you ever want to contribute? Well, guess what? It's much more possible now.

397
01:08:46.300 --> 01:09:08.690
Andrew Park: and a lot of people don't believe that and a lot of my job as a A leader has been to interact with my staff. You know. I'm giving them mandate. You got to use Gen. AI all day every day. But then, when I see that they're not doing it, then I will be like, well, you know. Okay, so you're not understanding how you could use it. But let me show you what you can do.

398
01:09:08.859 --> 01:09:13.350
Andrew Park: and so for you as a leader, if you're a leader of a company.

399
01:09:13.600 --> 01:09:35.000
Andrew Park: One of the things you really want to invest in is helping your people get through that stage of unbelief that, like what I could do design work. What I could write. Code. Yes, you can. But there's a lot of unbelief that gets in the way from them, really investing their human energies to do it, and sometimes you need to really handhold them a bit

400
01:09:35.269 --> 01:09:49.290
Andrew Park: or check in to make sure that they're trying it. And what you'll find is that for some of these people the light bulbs gonna go on where they're going to be like, I can't believe that I'm doing this work. This is amazing, you know. And and that's a real triumph

401
01:09:49.399 --> 01:09:51.469
Andrew Park: as a leader, if you can get there.

402
01:09:52.415 --> 01:09:53.229
Andrew Park: Okay.

403
01:09:53.734 --> 01:10:13.540
Andrew Park: Now, let's talk about usability testing for usability testing Pendo helps. Designers understand how people interact with their websites and apps it provides tools like heat maps which show where users are clicking or hovering the most. And it makes session recordings that replay user activity step by step. If you want to see what a user actually did

404
01:10:14.038 --> 01:10:32.280
Andrew Park: and these features are hugely useful. I'm a huge fan of Pendo. Pendo is amazing. It's got a big fan base, too. So so definitely, look into that now, hotjar offers similar usability testing tools including heat maps and session recording and things like that.

405
01:10:32.737 --> 01:10:51.700
Andrew Park: It also provides survey and feedback widgets that allow designers to collect direct input from users to help them better understand pain points and areas needing improvement. So both these tools aim to give designers clearer insights into user behavior, making it easier to improve the user experience with data driven decisions.

406
01:10:53.140 --> 01:11:18.309
Andrew Park: And then, earlier, I mentioned chat Gpts, custom Gpts, and how they allow domain experts import valuable knowledge from books, talks or documents to help teams learn through conversational guidance. And so for designers. If you're a business leader. You can use this to suck in all your internal knowledge. And then allow that to then help

407
01:11:18.310 --> 01:11:32.229
Andrew Park: be a resource to make sure that all your designers have easy access to sharing design principles, accessibility standards, brand guidelines in a consistent format.

408
01:11:35.560 --> 01:11:36.410
Andrew Park: Okay?

409
01:11:36.570 --> 01:11:40.140
Andrew Park: So the next thing is scrum masters.

410
01:11:40.300 --> 01:11:44.120
Andrew Park: So for facilitating daily scrums

411
01:11:44.290 --> 01:12:03.893
Andrew Park: on our teams we use stand up, Lee, to it's a stand up bot we use it to automate daily stand ups or daily scrums. So instead of requiring everybody to meet at the same time, you know, you can collect status updates asynchronously and prompt team members with customizable questions and

412
01:12:04.680 --> 01:12:25.079
Andrew Park: and then we also use pulse Point AI to provide automated insights into coding activity and project progress. And so this tool analyzes the actual source code, repository activity, all the commits and pull requests and generates daily reports that are then forwarded to all the team, and whoever else cares.

413
01:12:25.080 --> 01:12:43.779
Andrew Park: And so for scrum masters. These reports are especially valuable, because not only can they track their own team's progress, but they can also view reports from other teams, and that kind of cross team visibility is, really useful for managing dependencies between teams. And avoiding bottlenecks.

414
01:12:45.962 --> 01:13:03.320
Andrew Park: Now for facilitating sprint retrospectives, parable and retrium can help simplify the process by automating, meeting facilitation. These are not AI tools. So these are just useful software tools that I recommend looking into

415
01:13:03.560 --> 01:13:06.019
Andrew Park: and so

416
01:13:06.565 --> 01:13:14.610
Andrew Park: they help you. This automation helps you reduce a lot of the manual effort. The scrum masters typically invest in organizing and running retrospectives.

417
01:13:14.830 --> 01:13:15.820
Andrew Park: Now.

418
01:13:16.170 --> 01:13:42.649
Andrew Park: one of the really powerful things of using tools like this is that you can export that feedback from these tools and the summaries from paraball and retrium, and then you can process them with chat, gpt, and so then you can chat, gpt, and refine vague feedback or highlight recurring patterns across multiple retrospectives and turn general suggestions into clearer, more actionable steps.

419
01:13:42.680 --> 01:13:51.079
Andrew Park: And it can also generate polished summaries tailored for leadership or cross team sharing and improve visibility without extra manual effort.

420
01:13:51.380 --> 01:14:07.299
Andrew Park: So by combining these tools, scrum masters can not only streamline retrospectives, but also gain deeper insights into team dynamics and help them shift focus from facilitation to driving broader strategic impact across the organization.

421
01:14:10.100 --> 01:14:22.120
Andrew Park: In regard to facilitating other scrum ceremonies. Taskade, is a modern workflow management tool, and it's got AI at its core. And so

422
01:14:22.627 --> 01:14:38.849
Andrew Park: it'll help you automate task creation or prioritize backlog items and task progress based off of sprinkles. It can even generate summaries of your planning sessions and suggest backlog adjustments, cutting down on manual prep. Work. You often face as a scrum master.

423
01:14:39.110 --> 01:15:03.960
Andrew Park: And we also use sonics extensively for automated transcription. So it's a audio transcription tool. It's been a game changer for capturing key discussions during meetings. It provides fast, accurate transcriptions in over 40 languages, and it does an excellent job of identifying in individual speakers. That's a real real strength, and that makes it easy to maintain a clear, structured record of who said, What

424
01:15:04.389 --> 01:15:11.979
Andrew Park: and that's really helpful. When you're using Chat Gpt, to reflect on team discussions and decisions and follow ups.

425
01:15:12.110 --> 01:15:38.489
Andrew Park: And so when you combine sonics with Chat Gpt, since sonics identifies each speaker, chat, Gpt can help you break down a sprint review or retrospective transcript by summarizing key points from each person, and you can pinpoint blockers raised by certain team members. Raku suggested key ideas and generate summaries that clarify both team consensus and individual contributions.

426
01:15:39.003 --> 01:16:02.919
Andrew Park: And so Chat Gpt can use these sonic transcripts to analyze multiple transcripts over time revealing recurring blockers or patterns and feedback helping you to identify areas for continuous improvement with more precision. And of course, as you build up a large amount of these transcripts. You might want to use something that has a bigger context window and then chatgpt.

427
01:16:04.360 --> 01:16:27.500
Andrew Park: Okay. Now for agile coaching, professional development and domain knowledge, growth. Again, using custom Gpts can be a powerful tool. Take all the transcripts from Youtube talks or podcasts or workshops. That you like and put it all together and then you can access that information in a in a conversational way.

428
01:16:28.540 --> 01:16:33.880
Andrew Park: Okay, now for product owners, backlog management

429
01:16:34.350 --> 01:16:55.530
Andrew Park: for this. A lot of you are probably using Jira. And you probably noticed that Atlassian intelligence has been pushed on us to help summarize issues and improve documentation and things like that. And that kind of it works. Okay, right now. But but I'm sure it'll get much better much, you know, very quickly.

430
01:16:56.136 --> 01:17:07.789
Andrew Park: I also want to point out Aha! Aha has been a leading platform that combines backlog management idea, capture, roadmapping and progress, reporting in one place.

431
01:17:07.910 --> 01:17:17.330
Andrew Park: and it keeps strategy, backlog items and progress updates connected, making it easier to stay organized and keep stakeholders aligned. And it's not that expensive either.

432
01:17:17.560 --> 01:17:26.319
Andrew Park: and Chat Gpt can make both these tools even more effective on your day to day work. So with Jira it can help

433
01:17:26.460 --> 01:17:33.839
Andrew Park: clarify big issue descriptions or summarize ticket discussions and break down tasks into clear action items for the team

434
01:17:34.090 --> 01:17:46.560
Andrew Park: in Aha. You can use it to simplify complex roadmaps and identify patterns and user feedback giving you clearer insights for prioritization. And greatly reduce the manual effort to to get there.

435
01:17:46.670 --> 01:18:04.020
Andrew Park: Now, when it comes to user story drafting chatgpt is great. So I really recommend you to to use it and use it heavily. Jeff Sutherland is actually posted on this very topic and talks about how he does it. He's very strongly recommends it.

436
01:18:05.712 --> 01:18:22.730
Andrew Park: Now for team alignment. Our teams use pulse point AI, and this allows us to be able to have product owners to analyze the data from commits, pull requests and merges, giving product owners a clear view of how work is progressing against the sprint goals.

437
01:18:22.730 --> 01:18:36.990
Andrew Park: and this visibility reduces the need a lot for doing manual check-ins or frequent status meetings and the daily reports it generates makes it pretty easy to spot when progress may be stalled or work seems to be drifting from planned priorities.

438
01:18:42.120 --> 01:18:45.600
Andrew Park: Okay for customer engagement. I talked about Pendo.

439
01:18:46.249 --> 01:19:00.279
Andrew Park: Again, Pendo has tons of insights that you, as a product owner, can really really benefit from. If your if your company isn't using Pendo push for it. It's it's really amazing. You can thank me later.

440
01:19:01.770 --> 01:19:10.389
Andrew Park: Alright for professional development again. Just chat gpts custom Gpts can be really helpful

441
01:19:11.030 --> 01:19:27.259
Andrew Park: and to take it step further. We use sonics extensively to create transcripts from talk, talks and podcasts and webinars. And you can feed that into chat. Gpt to be able to build up your knowledge base for your custom. Gpt.

442
01:19:27.460 --> 01:19:28.315
Andrew Park: so

443
01:19:29.460 --> 01:19:30.840
Andrew Park: This approach

444
01:19:31.360 --> 01:19:45.049
Andrew Park: is, you'll see I've mentioned it for all roles, because it it's just universally applicable. And if you're a business leader again, you know, this is something that can really really help your whole organization, every role.

445
01:19:45.560 --> 01:19:48.630
Andrew Park: Now, one thing I want to say

446
01:19:49.070 --> 01:19:52.690
Andrew Park: is, I give advice to every product owner I meet

447
01:19:52.870 --> 01:19:56.380
Andrew Park: to aspire to become a full fledged product manager.

448
01:19:56.924 --> 01:20:16.449
Andrew Park: There is a big difference between product ownership and product management. And here I'm showing it. So this is the scope of a task. Let's say that product owners typically are responsible for. And then here's the additional things that product managers are responsible for.

449
01:20:16.620 --> 01:20:22.090
Andrew Park: And product management is really, actually a very, very young field.

450
01:20:22.752 --> 01:20:31.329
Andrew Park: And this is something that I'm I'm predicting is gonna become by far the most lucrative role on the entire product team.

451
01:20:31.570 --> 01:20:43.890
Andrew Park: And today's product managers will probably become tomorrow's Ceos. So if you imagine a person being able to do this, all of this versus doing this.

452
01:20:43.950 --> 01:21:10.430
Andrew Park: you could imagine how this person could be tapped to be a CEO of some some company and so the very good ones are going to command much bigger salaries than today. But these salaries for really good product managers are are rising very quickly Meta, just published that. They're up level senior positions for product managers. 1.2 million

453
01:21:10.838 --> 01:21:19.200
Andrew Park: we're gonna see this go to like 5 million, maybe even 10 million. In the future. It might be like sports contracts like, I'll give you a 10 year deal for 100 million bucks.

454
01:21:19.663 --> 01:21:27.339
Andrew Park: You know, when you're a driving product manager for a billion dollar sulfur product. Yeah, it was 10 million bucks a year.

455
01:21:27.630 --> 01:21:37.739
Andrew Park: So so this is something that, really, a lot of people should look into every young person that I meet that's going into college. I talked to them. Project management.

456
01:21:38.180 --> 01:21:39.050
Andrew Park: Okay?

457
01:21:39.310 --> 01:21:40.076
Andrew Park: So

458
01:21:41.830 --> 01:21:57.839
Andrew Park: there's if you want to learn about how Gen. AI has similar impact on all these roles? I only covered certain ones that I thought were relevant for this crowd. But yeah, we, you know, obviously the kind of discussion I had relates to every role.

459
01:21:59.400 --> 01:22:17.170
Andrew Park: And getting back to what I started with, what's my advice to you as form master. What's my advice to you, product owner? What's my advice to you, developer designer, whoever you are, project manager, leverage AI to become a cross functional performer.

460
01:22:17.958 --> 01:22:33.659
Andrew Park: Because the trend in the industry is that team structures are evolving to do more with less and so we're seeing a push towards a trend towards a cross, functional people replacing rigid roles.

461
01:22:34.308 --> 01:22:43.070
Andrew Park: Ai is a game changer for scaling your human talent, both horizontally, productivity and vertically. The sophistication of your work.

462
01:22:43.440 --> 01:22:51.760
Andrew Park: And you look into adopting AI tools to streamline current tasks, to boost your productivity

463
01:22:52.250 --> 01:22:54.200
Andrew Park: that does not obsolete you.

464
01:22:54.310 --> 01:23:22.439
Andrew Park: that makes you competitive in in performing your current role, but then it should free up a lot of time to do what? To become cross, functional, choose something else. To add to your skill set. So you can be part of this team that does more with less. And so, Master Gen. AI by using it all day every day, and when I 1st told my people that they should use it all day, every day, I said, I mean in your personal life.

465
01:23:22.490 --> 01:23:30.040
Andrew Park: really? Why? Because the more you do it in every situation, the more that you understand the capabilities and the limitations of it.

466
01:23:30.450 --> 01:23:37.289
Andrew Park: And so I use it for everything. I use it for cooking. I use it for my music, use it for everything.

467
01:23:37.901 --> 01:23:47.610
Andrew Park: and seek opportunities to contribute in adjacent roles. Alright, just if your company has not moved from this to this, they will

468
01:23:47.970 --> 01:24:11.269
Andrew Park: alright. So get ahead of the trend by figuring out what are you interested in? What are some adjacent areas that you might be interested in in playing and contributing in go try to get involved. Don't even ask for permission. Just just start contributing and prove to people that you have ability, and that you have interest.

469
01:24:11.900 --> 01:24:16.749
Andrew Park: and most finally, don't let your unbelief hold you back.

470
01:24:17.160 --> 01:24:18.806
Andrew Park: And so,

471
01:24:20.040 --> 01:24:25.089
Andrew Park: You know I'm able to manage my people to make sure that their unbelief doesn't hold them back.

472
01:24:25.390 --> 01:24:45.350
Andrew Park: But, if you're listening to this, you have to keep yourself accountable, and the possibilities of picking up new skills with Gen. AI is easy now compared to the past. It's super easy compared to the past, and this should be something that is very exciting and highly motivating for you.

473
01:24:46.740 --> 01:24:48.870
Andrew Park: Okay, just wanna

474
01:24:49.221 --> 01:25:16.910
Andrew Park: acknowledge our partners here. I've actually been able to interview some product managers from lucid, and I will say something about them. I'm really impressed with their company culture. It sounds like a great place to work. And then also, there's the agile 2025 conference coming up, I think what you'll see is that there's gonna be big changes happening in agile. And therefore you're gonna see big things happening this year in agile alliance.

475
01:25:17.302 --> 01:25:30.740
Andrew Park: It's no better time to get plugged into these talks and and other things that are happening in agile alliance, because there's big changes coming. And I think it's gonna be really important for you to stay on top of it.

476
01:25:31.360 --> 01:25:35.320
Andrew Park: Okay, here's how to engage me on social media.

477
01:25:36.060 --> 01:25:42.829
Andrew Park: a lot of the insights I shared with you. writing a book to collectively document them.

478
01:25:42.880 --> 01:25:59.230
Andrew Park: And I hope that that'll be coming out here this year. If you go to our website, you can see the outline. And if you want access to our coding technique training resources. You can email our support email address, and they'll get you access

479
01:25:59.680 --> 01:26:15.319
Andrew Park: and if you're interested in consulting. We do. We have consulting specialties. We help companies get self hosted. Gen. AI solutions in place so particularly important for companies that don't want to expose their data to public.

480
01:26:15.850 --> 01:26:45.540
Andrew Park: gen AI providers. And so we can install, on premise things for you that you completely control and train your people to take it from there. We also do safe optimization. So if you are dealing with safe programs, and you want to go a lot faster, that's our specialty. We can help you really supercharge that if you're dealing with a lot of technical debt on code bases, we're experts at helping people with that, too.

481
01:26:45.850 --> 01:26:48.199
Andrew Park: and that's the end.

482
01:26:51.700 --> 01:26:53.240
Mohan Vemulapalli: Well, thank you, Andrew.

483
01:26:54.050 --> 01:27:04.130
Mohan Vemulapalli: We have maybe time for just a few questions. And I did compile a list of that

484
01:27:05.100 --> 01:27:06.600
Mohan Vemulapalli: and let's see.

485
01:27:07.863 --> 01:27:13.540
Mohan Vemulapalli: I'm just going through this. I'm I'm thinking, maybe 3 questions. Anu, what? What do you think.

486
01:27:13.540 --> 01:27:14.970
Anuradha Goli(Anu): Yeah. No go ahead.

487
01:27:15.310 --> 01:27:39.139
Mohan Vemulapalli: Okay. So this is from cherise Livingston. And I think this is a good fundamental question. She asks, can someone explain software composer versus software development. And she goes on to say, I think the difference is, the composer is more design oriented and validation not sure. Okay, please advise. So, Andrew, please advise.

488
01:27:39.800 --> 01:28:09.429
Andrew Park: Sure. All right. So a software composer here. These circles here are representing where their human effort is intended to go. So for software developers. They spend some of their effort in design, not a lot. And then they spend most of their effort in implementation unit testing and a significant amount in integration system testing product release and then a little bit in release retrospectives.

489
01:28:09.560 --> 01:28:19.440
Andrew Park: Now, the software composer does not spend a lot of time in implementation unit testing. Because Gen. AI is a huge, huge accelerant there.

490
01:28:20.025 --> 01:28:31.670
Andrew Park: They spend less time in integration system testing and where they spend a lot more time is in design, so they get much better at design and craftsmanship.

491
01:28:32.078 --> 01:28:36.259
Andrew Park: And if people are interested I could do a future tech talk on that.

492
01:28:36.650 --> 01:29:00.599
Andrew Park: But then they also get involved in requirements and planning. So what is this? What I'm describing here is, I'm describing a product engineer. Okay? So an AI enabled product engineer is very much caring about the what and the why of the pro of the things that they're developing, not just the how.

493
01:29:00.630 --> 01:29:15.215
Andrew Park: And that's 1 of the very limiting things of software developers is that they're only thinking about how to do something. They're not thinking enough about the what and the why. And so leading companies that are having product engineering

494
01:29:15.580 --> 01:29:33.020
Andrew Park: professions where they're converting software engineers into product engineers. That's what software composers are. Basically just AI enabled product engineers so it, it's 1 of the things that allows you to get to a smaller team to do more with less.

495
01:29:33.300 --> 01:29:50.980
Andrew Park: And Meta is just one example. You'll see if you look at job postings that there's a number of Silicon Valley companies that are hiring product engineers. And if they're working for big tech, they're definitely using Gen. AI. So in that case, what I've drawn here is software composer is equivalent to their product. Engineer.

496
01:29:51.860 --> 01:30:03.971
Mohan Vemulapalli: Got it. So I don't think we have time to address all these issues, but I did want to throw out. We've had a number of questions about how the these new tools and practices

497
01:30:04.320 --> 01:30:25.170
Mohan Vemulapalli: relate to traditional scrum roles and really scrum roles, not not say Xp or other other flavors of agile. So I thought that was interesting. Maybe something for you to think about in future, Andrew, there are 2 more questions I want to throw out very quickly, and we're going to hit the limit. So

498
01:30:25.170 --> 01:30:46.189
Mohan Vemulapalli: Claire Lynch had an interesting question. That is, what about the environmental cost of using AI all day every day. I know this has been an issue, say with blockchain, and you know other. You know, other intense, you know, intense resource using systems. So I'm just wondering what you think about that.

499
01:30:46.760 --> 01:31:06.750
Andrew Park: Yeah, I think that that's outside my expertise. Yeah, I think that. There's there's that's an issue that. I think the political leaders of the world have to deal with and come to. Some conclusions on. Another thing is AI safety. Okay. Ai, safety is a huge thing.

500
01:31:07.085 --> 01:31:18.499
Andrew Park: I'm not talking about that. That's kind of outside of my purview in my job. But it is something very much that politicians and and leaders need to. They need to deal with that.

501
01:31:19.112 --> 01:31:33.070
Andrew Park: So there's things that are within your control. There's things within your life that that I'm trying. I'm trying to talk to certain people that want to know

502
01:31:33.230 --> 01:31:54.749
Andrew Park: what's the impact of Gen. AI on you as a worker and your employability in the future, how to set yourself up for success in the future and to make sure that you're not a statistic that isn't employed anymore. Okay? And hopefully, you find from here that I'm laying out some wisdom for you of of what to do and some encouragement to do it immediately.

503
01:31:56.130 --> 01:31:59.596
Mohan Vemulapalli: Very good. Know the guardrails and respect them.

504
01:32:00.430 --> 01:32:12.410
Mohan Vemulapalli: thank you. And then this last one is from Dave Brodko, and I'm I'm picking some of the more difficult ones. So since you're

505
01:32:13.130 --> 01:32:34.449
Mohan Vemulapalli: your presentation was so comprehensive it actually answered many of the questions I'd anticipate. But this one, I think, is interesting. If if AI means you don't need as many devs overall and senior Devs are more useful. How do Junior Devs get the experience to become Senior Devs that will get hired to work with AI. Now we do think you covered part of that. But perhaps you could elaborate.

506
01:32:34.950 --> 01:32:38.039
Andrew Park: Yeah. Well, number one, it's harder on them.

507
01:32:38.440 --> 01:32:51.929
Andrew Park: This is the 1st time in the history of my career that I've met graduates. Computer science graduates from top schools that cannot find a job.

508
01:32:52.080 --> 01:32:53.200
Andrew Park: So

509
01:32:53.460 --> 01:33:08.980
Andrew Park: it sucks for that. Okay, so there's nothing. Past that. Yes, it's just hard for them right now. But this is something that it's temporary. Okay, so what I think is gonna happen is another prediction I'm gonna make is that

510
01:33:09.900 --> 01:33:29.069
Andrew Park: In addition to all these layoffs that are going to happen at many, many companies, because they're going to find that they're not going to need as many people as what they have right now. There's going to be a lot of layoffs. And what is going to happen after that? You're going to have a lot of excess capacity. And with that excess capacity you're going to find a lot of people realizing that they could have

511
01:33:29.200 --> 01:33:34.180
Andrew Park: a 3 person company, a 2 person company with one developer.

512
01:33:34.280 --> 01:34:01.740
Andrew Park: and it could even be a junior developer. And they could do a lot of stuff, you know. And and so at Openai. Sam Altman and his colleagues have been making bets about the year that they're gonna see the 1st one person 1 billion dollar company 1 billion, as in B right? And and so this, there's gonna be a plethora of tiny little companies.

513
01:34:01.960 --> 01:34:24.980
Andrew Park: And and this is why I push people into product management and sales business. There's gonna be a lot more jobs in business and sales and product product management than there are today. Because it's gonna be very viable to have very, very small companies that actually do pretty big things.

514
01:34:25.449 --> 01:34:40.610
Andrew Park: And the rise of AI agents, and purposely didn't talk a lot about AI agents, because that could be a whole session by itself. But but that is going to greatly accelerate this reality of being able to have micro companies that actually do big things.

515
01:34:40.700 --> 01:34:56.780
Andrew Park: And so there's going to be this intermediate period with a lot of pain for a lot of people. But then, you know, it creates the the ground for new things to grow. And I'm actually very optimistic about the future.

516
01:34:57.520 --> 01:35:00.740
Mohan Vemulapalli: I see. So the so the glass is half full.

517
01:35:01.210 --> 01:35:02.729
Andrew Park: Yeah, I view it that way.

518
01:35:03.950 --> 01:35:09.145
Mohan Vemulapalli: Okay, I think we're probably at time, unless

519
01:35:11.376 --> 01:35:14.500
Mohan Vemulapalli: Teresa or Joe want to continue.

520
01:35:16.100 --> 01:35:19.469
Agile Alliance: We can continue if you want to continue. That's fine.

521
01:35:20.510 --> 01:35:21.030
Andrew Park: But.

522
01:35:22.320 --> 01:35:26.209
Agile Alliance: I don't know if people have time, we can certainly stay on.

523
01:35:28.370 --> 01:35:29.290
Mohan Vemulapalli: Okay?

524
01:35:29.410 --> 01:35:41.770
Mohan Vemulapalli: Well, I want to take a moment also to thank our Agile Alliance's annual partners, lucid and Getsoft and Joe did post the links before. So check that out.

525
01:35:42.140 --> 01:35:47.939
Mohan Vemulapalli: And Andrew, any more questions you up for that.

526
01:35:48.680 --> 01:35:49.580
Andrew Park: Sure.

527
01:35:50.320 --> 01:35:51.150
Mohan Vemulapalli: Okay.

528
01:35:52.724 --> 01:35:55.809
Mohan Vemulapalli: Anu, did you wanna jump in.

529
01:35:55.810 --> 01:36:03.180
Anuradha Goli(Anu): Yeah, yeah, I just wanted to ask a question that started at the initial conversation. Right? This is a question from dhruv.

530
01:36:03.716 --> 01:36:15.899
Anuradha Goli(Anu): He said, like, focus, Andrew, you mentioned focus on the business outcomes, but not on the metrics. Right? So his question is like, how do we measure those business outcomes.

531
01:36:16.760 --> 01:36:19.920
Andrew Park: I'm sorry your volume is very low. Could you repeat that louder.

532
01:36:20.410 --> 01:36:33.699
Anuradha Goli(Anu): Oh, sorry this is a question from Drew. So it's like when you stated a statement saying like, focus on the business outcomes, but not on the metrics. So his question was like, How do we measure those business outcomes?

533
01:36:35.425 --> 01:36:36.080
Andrew Park: Yeah.

534
01:36:36.677 --> 01:36:57.419
Andrew Park: business outcomes are things that sometimes are not easy to measure. Sometimes they are easy to measure so as an example. And this is where you want to really ally yourself, like, I said with product and with sales. So you know, there's some organizations that have moved to the product operating model.

535
01:36:57.880 --> 01:37:02.330
Andrew Park: In that case there's a lot of product managers. Those are the people you want to talk to.

536
01:37:02.868 --> 01:37:13.689
Andrew Park: But many organizations, even more organizations are still in a sales led model where you don't really have a lot of product managers. So then you want to talk with the salespeople

537
01:37:14.583 --> 01:37:16.649
Andrew Park: when you go to them

538
01:37:16.850 --> 01:37:24.919
Andrew Park: they will have ideas of what they want. And so one example could be that maybe a product manager would say.

539
01:37:25.593 --> 01:37:36.349
Andrew Park: Oh, one of the outcomes that I want is, I want to double the amount of active users. By next quarter. Okay, well, that's measurable.

540
01:37:36.820 --> 01:37:55.700
Andrew Park: And or sometimes it's not measurable as far as something that you could put something in place to say. Oh, we got there, but you could still qualitatively ask a product manager or salesperson. What is what? What are your top goals? What do you? What do you hope to be able to see?

541
01:37:56.110 --> 01:38:04.090
Andrew Park: And that's really all you need to do is you need to understand the mind of the people that are driving business results.

542
01:38:04.260 --> 01:38:27.170
Andrew Park: And the better that you understand that more that you can do. As a person to be a force in making that happen. So there's gonna be layoffs. But the people that are producing results are the ones that are gonna have a much, much greater probability of not getting laid off.

543
01:38:27.320 --> 01:38:28.059
Andrew Park: And

544
01:38:29.070 --> 01:38:43.670
Andrew Park: so you need to really understand, what the business is trying to get done. And people in product, people in sales are the ones that have a pretty darn good idea of of what they're shooting at.

545
01:38:43.770 --> 01:39:09.069
Andrew Park: And usually in most organizations. Once you get to the agile teams, they really have a scant understanding what that is, and that is something that really is gonna be changing in the future. And when I talk about agile turning into something that encompasses product management and design I think that that kind of trend that I'm seeing happening is because.

546
01:39:09.545 --> 01:39:36.359
Andrew Park: the agility and the metrics that that we very much popularized these these years have not really made the rubber hit the road in many situations, and that's why that's why Ceos don't want to hear agile consultants talk about agile or scrum and things like that. But but they're talking to people about business agility. So so they're they're the

547
01:39:36.430 --> 01:39:41.149
Andrew Park: attractiveness of agility has not diminished at all. It's just

548
01:39:41.717 --> 01:39:45.299
Andrew Park: the scope of it, I think, has increased.

549
01:39:46.080 --> 01:40:11.230
Anuradha Goli(Anu): Yeah, okay, yeah, no. That makes sense. Actually, do you have time for one more question. So I, probably, yeah, okay, so I know we you have mentioned plenty of tools in this, and lot of things that can be replaced with the general AI tools like, how is this aligning with this agile practices of like process and tools over, like, you know, interactions and individuals.

550
01:40:11.340 --> 01:40:13.849
Anuradha Goli(Anu): How is it aligning, you know.

551
01:40:14.220 --> 01:40:40.529
Andrew Park: Yes, I think that there's I've had a conversation with someone. When I talked about I said, Hey, you should use a stand up bot and and this was an agile coach, and they're like, No, I. And I can't agree with that, Andrew, because that's putting a tool instead of individuals and interactions. And and this is where you've got to put on common sense.

552
01:40:40.820 --> 01:40:46.560
Andrew Park: Okay? Because what really matters is results.

553
01:40:46.630 --> 01:41:16.289
Andrew Park: What really matters is business agility. And when the agile manifesto came out, and all the principles, those were all things that in the waterfall world were hopeful about trying to refocus people to deliver more results rather than Hey, let's deliver lots of artifacts underwater. Remember, there's tons of artifacts being delivered, lots of effort to do that. But there wasn't really good results, and everyone knew it.

554
01:41:16.781 --> 01:41:32.709
Andrew Park: Now, you fast forward to today, and you see the same disdain that, hey? I'm not getting the results I want, and I need to go faster. And 2 week sprints are way too slow, and things like like really common things I'm hearing and and what

555
01:41:32.920 --> 01:41:38.789
Andrew Park: you if you just look at those statements as religious beliefs.

556
01:41:39.280 --> 01:41:54.389
Andrew Park: it's gonna hold you back from making the right judgments of what to do. Okay, if you can automate a lot of the things that are causing all these meetings and all these distractions, guess what happens? Productivity goes up.

557
01:41:54.570 --> 01:42:02.359
Andrew Park: And now, when productivity goes up due to less distractions, less meetings, what can you do? You can get more done.

558
01:42:02.660 --> 01:42:15.699
Andrew Park: You know. What will happen is that there still is gonna be just as much individuals and interactions. It's just completely different types of individuals and interactions. And they're much, much more productive towards business agility.

559
01:42:16.070 --> 01:42:26.390
Andrew Park: So so this is something where I've I've gotten into a lot of conversations about this very thing, and maybe that's worth the tech. Doc.

560
01:42:27.060 --> 01:42:34.930
Anuradha Goli(Anu): Yeah, I think I think this is very insightful. And we just get to learn so many things and so many tools. But this, thank you so much, Andrew, for this.

561
01:42:35.630 --> 01:42:36.170
Andrew Park: Okay.

562
01:42:36.440 --> 01:42:46.790
Mohan Vemulapalli: Very much. My last question to you is, look into your crystal ball, and tell us if you what do you think is on the horizon for AI.

563
01:42:49.570 --> 01:43:03.919
Andrew Park: That's really hard to say right? Because I think that the near term horizon is very clear the long term horizon is very murky. So the near term horizon is AI agents.

564
01:43:04.430 --> 01:43:14.840
Andrew Park: So if you haven't heard about AI agents, get on Youtube, learn what this is, or or you just go to chat Gpt. Say, what are AI agents and explain it to a way that I can understand.

565
01:43:15.439 --> 01:43:20.190
Andrew Park: This is, gonna be the year that AI agents break out into the scene.

566
01:43:20.330 --> 01:43:26.239
Andrew Park: and and everybody a year from now is gonna understand what this is, because it's gonna be everywhere.

567
01:43:26.972 --> 01:43:36.230
Andrew Park: And the reason why is because if you understand what I shared with you about the strengths and the limitations

568
01:43:36.280 --> 01:43:58.149
Andrew Park: of AI, those limitations are what AI agents are kind of a trying to solve, you know, by instead of making Gen. AI, which is very general. If you train up a really specific AI for a specific task. It can produce a lot better results.

569
01:43:58.523 --> 01:44:24.179
Andrew Park: With much less computational power than maybe it would required it for a general model. And so you could think of these agents as being very, very specialized trained things. And but they can do only like this one thing or this 2 things. And so it's very narrow. But that's okay. Because if you have layers above it to put them all together, then oh, maybe you can. Maybe you can do something right?

570
01:44:24.180 --> 01:44:34.119
Andrew Park: Something big. So I think that it'll still have limitations. I still don't believe for a second it's gonna obsolete. All the humans in the world.

571
01:44:34.890 --> 01:44:44.800
Andrew Park: I think again, it'll it'll be a tool that will allow us to automate a lot of things and and be able to change where we focus as as humans.

572
01:44:44.920 --> 01:45:06.279
Andrew Park: But yeah, AI agents is the near term thing that's pretty clear as far as the longer term. Who knows? You know, even Sam Altman doesn't know. He'll speculate, but he'll be very clear that he's just speculating and and he'll also be very clear that he was very wrong in his speculations. 3 years ago. So

573
01:45:06.500 --> 01:45:23.210
Andrew Park: so what we do know for sure is that AI is bringing massive changes. And what I tell my clients is that AI is like the industrial revolution, times 55 0.

574
01:45:23.540 --> 01:45:30.640
Andrew Park: It's going to impact every corner of the world just about every human occupation in the world.

575
01:45:30.940 --> 01:45:45.869
Andrew Park: And and so it's gonna have 50 times the transformative effect. But it's gonna come so much faster. You know, it took decades and decades and decades for the industrial revolution to really play out.

576
01:45:46.330 --> 01:45:51.140
Andrew Park: And AI is completely different.

577
01:45:51.953 --> 01:46:07.700
Andrew Park: And then, if quantum computing becomes real, you know, and we don't know. But if it does oh, my, gosh, all right. Now, that's that's something that's gonna you know. Be 100 x of whatever the speed was before. So

578
01:46:08.110 --> 01:46:16.679
Andrew Park: so who knows? One thing is very clear is that the world is changing faster than it ever has in human history.

579
01:46:17.040 --> 01:46:35.919
Andrew Park: People are afraid of change. Don't be afraid of change. You can't stop it, by the way, all right. But what you can do is try to understand it and try to adapt to it. And isn't that the whole foundation of agility? Right? We all believe that we all know that.

580
01:46:36.320 --> 01:46:57.439
Andrew Park: you know it's the people who evolve that survive, and I would say, not only survive but thrive. And so what I'm hoping is that I've given you some motivation and some general wisdom of how not to just survive as an you know, an employable person, but hopefully to thrive.

581
01:46:59.460 --> 01:47:03.730
Mohan Vemulapalli: Wonderful. Thank thank you. You've given us quite a lot to think about.

582
01:47:06.040 --> 01:47:07.306
Andrew Park: You're welcome.

583
01:47:08.170 --> 01:47:09.080
Mohan Vemulapalli: Thank you.