Opening Remarks — Day 2

Host Intro (Gene Kim)

Good morning. Welcome to day two of the Enterprise AI Summit.

By a show of applause, how was everyone's evening yesterday after the talks were over?

Awesome. I am so glad you're here. We have such an amazing day of programming and talks for you. You're going to hear more amazing experience reports and expert talks on things we need to know to make good decisions.

I want to share just a couple brief remarks, and I have to try to give the speakers this morning a little more time. And it's really around three things: open questions, the value of experience reports, and the value of conferences and journals, and how we support the time of massive distributed learning.

The first slide is about open questions. So many of you know — imagine a slide of the DORA metrics. So you might imagine that this is something I got to work on with Jez Humble and Dr. Nicole Forsgren from 2013 to 2019. It's one of the things I'm most professionally proud of. It's a cross-population report that spans 60,000 respondents over six years, and gave us some of the metrics that many of us know in terms of what high performance looks like. But I think there are so many questions now.

I think the DORA metrics still apply, but they just so incompletely capture the magic that we feel with agentic coding. I was a part of the DORA effort last year. I've read all the things from DX about their studies, and they're saying that there's a 20% productivity increase. And I think that's such an incomplete expression of the magic that we feel. We know that it's possible to get two orders of magnitude more speed in terms of delivering value to customers. And so I think that's one interesting anomaly.

And what my personal experience with this is — some of you may know that Swyx, who was mentioned yesterday by Mik Kersten, runs the "Latent Space" podcast and runs the AI Engineering Summit. And he had this amazing contest. It was "kill my SaaS in one weekend." He offered a $10,000 prize to kill his $4,000 SaaS and to do it in a weekend. And I was interested in it, but on Saturday morning, when I actually found out which SaaS he wanted to kill — it was his call for papers software.

So we've run 24 conferences over 12 years on five of the worst software packages ever written. And for those of you on the program committee, you know how hateful some of these software projects are, right? And in fact, for April, it was actually such a pain in the butt I just used Basecamp. So welcome to 2007.

And so I jumped in with a passion. I've had 12 years to think about what the ideal CFP software would be. And I hesitate to bring this up because I was up till 5:00 a.m. for the next seven days, because I was so passionate about it. And thus was born Curtain Call. That was in production three days later.

And there were so many things I learned. I ended up with five $200 subscription accounts from Anthropic and OpenAI. 68 people completed it. And it just, I think, shows that the barrier to entry for many offerings is quite low, especially for software that you hate.

And I'll also say that, boy, I say I love it — I'm not sure if Anne Perry, who runs conference operations, loves it. But yeah, I think it totally has changed how we are planning and organizing conferences. So it actually knocked out three pieces of software that we use, two of which we love. But just having them all integrated in one space was just really, really powerful. And so I suggest this as one small experience report.

And as I started having my own opinions on some questions that were brought up yesterday — for example, who adopts AI more, juniors or seniors? And as Kent Beck famously said, the safest answer to most questions these days is, "I don't know." More tech debt or less? Don't know. How do we keep deploying code even though we don't feel completely in control? Is that okay? Don't know. Is Claude safe enough to use in production? Don't know.

And so I think, despite all of these things that we don't know, what's interesting about this community is that you are forging ahead, pushing the frontiers of performance, bringing this crazy new technology into your organizations to help move the needle on the most important things in your organization. And I think one of the things — I think the value that this community creates is creating guidance. And so you heard from Randy Shoup and Michelle Gill yesterday presenting on some collective guidance from some of the people I admire most in terms of how do we use this crazy new technology in a way that's responsible? How do we go back to the principles that we know still remain true, regardless of whether the code was written by AI or by humans? And I think this is so powerful.

So I want to give you just a sneak peek behind where these guidance papers — the forum guidance papers — come from, of which we've done over 150 in the last 10 years. Every year, we invite for three days some of the people I admire most, and we get together to write papers. So my boss and wife, Marguerite Kim, she says, "It's Gene's ultimate slumber party, where we pick the biggest problems in the industry, and we spend three days writing." And it's just an amazing experience because you're hanging out with the best in the game, you're working on problems that you're passionate about, and it's just one of the most intense learning experiences that I ever experienced in my career.

And so when we talk about conditions where the world had no idea what was going on, it reminds me of the scientific revolutions over the years. In the mid-1800s, there were some actually genuine, very troubling questions about the limits of Newtonian science. Things seem to work in the large, but as you get into smaller scales — both time and space — everything starts breaking down, and people are really worried about some crazy things. One of the questions was, is light a particle or a wave? And this actually caused great existential despair in physics. And so there was the Michelson-Morley experiment where they actually proved that it's both — which is crazy, preposterous, it could hardly be believed. And so it led to something called the Solvay Conference in 1927 that led to the greatest minds in science assembling.

And it just demonstrates that the breakthroughs in almost any domain never come from a lone genius. They come from a scene of like-minded spirits who have a common way of communicating, where knowledge is rapidly disseminated and shared. And I think that's what happened here with the DevOps movement, and it's happening here again in how we adopt AI in large organizations.

So I was very inspired by Art Smalley's remarks where he said, "Oh yeah, Toyota — they had these journals that were devoured by people in machining tooling design for engines." And it was inspiring to me, like, oh, I think in any kind of community, you find a place where people can share their learnings, experience reports large and small.

And so the forum paper — that's a huge heavy lift. It takes six, seven months. Many start, fewer finish. It's that tough of a lift. And I'm just wondering, how can we lower that barrier of entry so that anyone with an experience report, even as small as my experience report with Curtain Call — can we write that up where we can share that with this community of leaders who we trust to share experience reports with.

And so now I think that is going to be one of the goals with SELF: when you register, you're now part of a community of people where — if you don't know them, someone you know knows them. So there's these concentric rings of mutual trust, and we can share stories I wouldn't normally be able to share in a venue like this. So that is a part of the learning inspiration I got from day one.

So speaking of sharing learnings — believe it or not, most of the presentations from yesterday are actually already online. If you go to videos.itrevolution.com, all those videos are actually available through some incredible vibe coding and some ingenuity between the AV team and the IT Rev team, including Alex. So if you're interested in sharing the videos that you saw with people who would be interested, they're already available to you.

By the way, for the last 24 conferences — 10 years — our average is two weeks. One time we were able to get slides up and presentations up within a day. So the fact that we did it this time, I'm so happy about.

So with that, before we move into the talk, I'd like to turn it over to my co-emcee, Andy Ramirez.

Andy Ramirez

Thank you. I just wanted to say thank you again for having us here.

Right. Speaking of conferences — we can now announce that the next Enterprise AI Summit is now announced, and it's on May 5th and 6th in Portland, Oregon. Our hometown. So go Portland Trail Blazers, and save the date. Thank you.

Awesome. And I'm really looking forward to being there — I will absolutely be there. And again, I want to make sure you all take your time to go visit the sponsors who've made this possible. It's not just GitLab. We've got Coder and Digital.ai, Equal Experts, Liatrio, Lucid, Delphix by Perforce, PlanDeck, Planview, Quotient, Veracode, and Exalt. And it really is important that we give them the love back so that we can continue doing this. And again, thank you all very much for having us. Make sure you get on the Wi-Fi, and I appreciate y'all.

Gene Kim

Thank you, Andy.