AI Engineer Summit 2023
Announcing the AI Engineer Network
Read the talk
Network: Matching AI Engineers Around the Problems They Need to Solve
At the inaugural AI Engineer Summit, Network adds LLM-based attendee matching to an event app, using the problem in a participant’s profile as the starting point for a connection.
From a talk by Benjamin Dunphy
An emerging ecosystem in the room
Who has gathered to build the emerging field of AI engineering? The AI Engineer Summit 2023 opens with two days of talks from founders and engineers, with Swyx due to set the technical context. Benjamin Dunphy begins by looking around the room. AutoGPT supplies an immediate example of how quickly the ecosystem is changing: a few months after being an open-source project, it is the summit’s presenting sponsor. Dunphy calls it the fastest-growing open-source project in history, without specifying a growth metric or comparison period, and points ahead to announcements from Toran.
Supabase brings a different combination: playful social media and product launches, alongside databases whose ease and trustworthiness Dunphy associates with being open source. He says the company became a diamond sponsor despite its policy against event sponsorships, then asks Paul and Ant to let him know how the policy change works out.
Fixie makes the possibilities more concrete. Dunphy wants to ask its booth sidekick whether it can see what he is wearing and recommend wardrobe updates; he nicknames the sidekick Derek. The visual interaction remains hypothetical. If vision is not available yet, he offers Matt and Ben an introduction to Logan Kilpatrick for a possible Vision API preview. This is an October 2023 prospect: OpenAI’s public vision API announcement followed in November with GPT-4 Turbo image inputs through gpt-4-vision-preview.
Suggest correction
This note stays in this page until you copy or download it. Nothing is submitted; reloading clears the draft.
From investment to conversations
Dunphy credits Microsoft’s $13 billion investment in OpenAI with helping commercialize the movement. That figure refers to cumulative funding across 2019–23, as described in a later Microsoft-hosted report; the contemporaneous partnership announcement described a multiyear, multibillion-dollar investment. For founders in the room, the immediate opportunity is a conversation with Microsoft for Startups representatives in the Carmel Room, past the elevators behind the audience. GitHub’s neighboring booth provides a destination for Copilot questions. These are introductions to teams, rather than a promise of startup funding.
Cloudflare gets a belated birthday greeting: at 13, Dunphy jokes, it will be going through some changes. Its booth is next to Microsoft in Carmel, across the elevators. He then recognizes Alessio Fanelli, the venture-capital half of the Latent Space Podcast, whom he describes as no less technical than Swyx. Perhaps the talks will draw Fanelli back into founding—though Dunphy suggests a software startup this time, leaving hardware to the experts.
Suggest correction
This note stays in this page until you copy or download it. Nothing is submitted; reloading clears the draft.
Can already useful conversations become more useful?
Dunphy describes the inaugural summit’s audience as 500 selected attendees. They are experienced software engineers experimenting with production AI, or founders building AI-enhanced or AI-native applications and companies. That shared experience matters when attendees meet: the person beside them is likely working through related practical questions. He also recognizes livestream viewers and points them toward the next day’s opening keynote for announcements about expanding participation in 2024.
Conferences already do some of the work of finding relevant people. They bring experts from around the world together around a narrow subject, making many conversations useful before anyone has arranged an introduction. The next question is whether that concentration of expertise can produce better connections still. Shared interests make a conversation relevant; a stated problem gives a connection a purpose.
Suggest correction
This note stays in this page until you copy or download it. Nothing is submitted; reloading clears the draft.
A profile problem becomes the basis for a match
The Network application starts with the practical information attendees need on their phones: the conference schedule, the full expo partner list, other event information, and recommended restaurants, bars, and cafés within walking distance. On top of that event companion, Dunphy announces AI-enhanced attendee matching.
The feature is called the Generative Matching Algorithm, or GMA. Dunphy gives the name a plain explanation: it uses LLMs to match attendees with the right people. The announced interaction is straightforward:
- Enter the problem you want help solving in your Network profile.
- Let GMA use that stated problem as the basis for matching.
- Connect with people the system identifies as able to help.
The problem field is the substantive input described here. The announcement does not specify a model, prompt, ranking method, or evaluation of match quality.
Dunphy directs attendees to the conference’s Network download page. At launch, matching is limited to people attending this summit in person. That restriction defines the available pool for the announced feature, even though the app’s eventual ambitions extend beyond this event.
Suggest correction
This note stays in this page until you copy or download it. Nothing is submitted; reloading clears the draft.
Opening the app and matching algorithm to other events
The final announcement expands the scope: Dunphy announces that both the event app and the matching algorithm are being open sourced. The goal is a simple, powerful mobile experience for attendees at events around the world, with the matching component helping those attendees find one another.
The implementation credits separate the application, matching, and infrastructure responsibilities:
- Mobile app: Simon Sturmer, credited as lead engineer.
- Generative Matching Algorithm: Suizette Steller.
- Authentication: Descope.
- Database and vectors: Supabase and pgvector.
Dunphy invites attendees to talk with the partners outside about pgvector. Its inclusion identifies part of the infrastructure, but the announcement does not explain how vectors participate in GMA’s matching decisions.
After thanking the developers and partners, Dunphy announces that the repository is available that day through the Network landing page and GitHub. The closing slide supplies the destination: aiDotEngineer/Network. The release invitation carries the idea beyond the summit: other events should be able to provide the mobile experience and help participants connect around the problems they bring.
Suggest correction
This note stays in this page until you copy or download it. Nothing is submitted; reloading clears the draft.
Resources
From the talk
Source repository for the mobile event application and its attendee-matching project.
Historical event page with the program, speakers, workshops, venue, and presenting organizations.
Postgres extension for storing vectors and performing similarity searches, with installation and SQL examples.
Further reading
Current guide to enabling pgvector and storing embeddings in a Supabase database.
Current starting guide for adding authentication with Descope Flows or SDKs.
Updates since the talk
DevDay announcement introducing GPT-4 Turbo image inputs through the Chat Completions API.
Read the complete timestamped transcript
- 0:00
[upbeat music] Engineers, founders, sponsors, partners, colleagues, and friends, welcome to the AI Engineer Summit
- 0:24
2023. I am- [cheering] Yeah. [laughs] [clapping] Yeah. I am deeply honored to have the opportunity to host you all at this event with my co-host, Swyx.
- 0:39
And I'm especially delighted to kick off the presentation portion of the summit. We've curated two days of stage talk content for you from some of the top founders and engineers in this versioning new industry of AI engineering.
- 0:55
In just a few moments, Swyx is going to come on stage to help set the context for those talks. But for my time on stage, I really wanna know who is here.
- 1:07
So AutoGPT is here.
- 1:09
Whoo. [clapping]
- 1:13
A few months ago, they were just an open source project. Mind you, the fastest growing open source project in history, but just an open source project. Now, they're the presenting sponsor of the AI Engineer Summit.
- 1:29
So needless to say, we can expect some big announcements when Toran takes the stage in a few moments.
- 1:36
Let's see who else is here.
- 1:37
Whoo. [clapping]
- 1:37
Supabase. Supabase is here. The social media experts who make product launches fun,
- 1:46
and who make databases both easy and trustworthy, uh, given that they're open source. They're a diamond sponsor this year-
- 1:55
Whoo
- 1:55
... and we're honored that they broke their no event sponsorships policy for this event. [laughs] So Paul, Ant, uh, let me know how that policy change works out for you. [laughs]
- 2:06
Fixie is here.
- 2:08
Whoo. [clapping]
- 2:08
Where's Fixie? I'm looking forward to s- asking their booth sidekick if they can see what I'm wearing yet, and if they can recommend any updates to my wardrobe. We'll call this sidekick Derek.
- 2:22
Uh, and if Vision is not in the cards yet, Matt and Ben, uh, I can make an intro to Logan Kilpatrick, and perhaps we can get a preview of that Vision API.
- 2:31
Microsoft is here, the company who helped to commercialize this movement by pumping $[REDACTED:age] billion into OpenAI. [laughs]
- 2:40
Whoo.
- 2:40
And many of you might be thinking, "Well, how can I get some of that gravy for my startup?" [laughs] Well, head over to their booth in the Carmel Room, just past the elevators behind you, and have a chat.
- 2:52
They have representatives from Microsoft for Startups who can help to open those doors for you. And if you have any questions for the Copilot team, well, Microsoft's booth is right next to GitHub.
- 3:04
Oh, and happy belated birthday, [REDACTED:username]- Flare, who is here.
- 3:08
Whoo. [clapping]
- 3:09
Round of applause. They have, uh, [laughs] Cloudflare, you are now [REDACTED:age], so that means you're gonna be going through some changes. [laughs]
- 3:18
But don't worry, there's plenty of folks who will be delighted to talk to you about those changes at your booth, just next to Microsoft and Carmel, across the elevators.
- 3:27
Alessio Fanelli is here.
- 3:29
Whoo. [clapping]
- 3:31
The VC half of the Latent Space Podcast, but no less technical than Swyx.
- 3:38
Perhaps after watching some of today's talks, we can convince him to come back to the founder world a little bit easier, right? But this time, Alessio, let's get you a software startup and leave the hardware to the experts. [laughs]
- 3:52
Some of you got that joke. [laughs] And who else is here? Well, of course, you. You're one of 500 people who were selected to attend the inaugural AI Engineer Summit.
- 4:05
So give yourselves a round of applause. [clapping] This means that you're not only an experienced software engineer, but that you're actively experimenting with shipping to production, or have founded an AI enhanced or AI native apps and companies.
- 4:23
So I want you to remember this when you're interacting with one another. But we also wanna recognize the many people who are not here today, but are watching the live stream.
- 4:32
We thank you and appreciate you, and we hope that you enjoy the content on the live stream. And be sure to watch tomorrow's opening keynote address for some exciting announcements for 2024, which will expand our participation limits.
- 4:46
But for those of you who are here, we hope that you've been enjoying your conversations with other attendees so far. One of the reasons that I really love conferences is that it brings together experts from around the world who are passionate about a singular niche subject.
- 5:04
And so most of your conversations are already high signal. But is there a way we can actually improve this?
- 5:12
Many of you have downloaded our conference mobile app, Network, and have been enjoying its use. It includes the conference schedule, a full list of all of our expo partners, recommended restaurants, bars, and cafes within walking distance of the venue, and other event information.
- 5:32
But in addition to these features, we're pleased to announce AI-enhanced matching using our Generative Matching Algorithm, which is a fancy way of saying we use LLMs to match you with the right people.
- 5:47
We can match you with people who help to solve your stated problems.
- 5:52
We all have problems we need to solve. That's why in your Network profile, you can literally tell us what problem you're looking to have solved, and our GMA will connect you with the right people.
- 6:07
We call this app Network. You can download the app today at ai.engineer/network. The matching is currently limited to in-person attendees of this summit only, but we have one more big announcement that we'd like to make.
- 6:26
We are pleased to announce that we are open sourcing Network, both the event app and the matching algorithm. Our goal is to provide a simple yet powerful mobile event experience for attendees at any event around the world.
- 6:40
And we hope that the matching algorithm can better assist to connect people at those events.
- 6:46
And we'd like to thank our lead engineer, Simon Sturmer, for the mobile app-
- 6:50
Whoo
- 6:50
... and Suizette Steller for our Generative Matching Algorithm, and our infrastructure partners, Dscope for Auth, and Supabase for database, and pgvector. So go talk to them about pgvector outside.
- 7:01
So a round of applause for all of our- [clapping] ... developers, partners
- 7:07
for this app. So you can access the repo today at ai.engineer/network, or github.com/ai.engineer/network.
- 7:23
So with that, I'd love to welcome our first speaker to the stage. He's the co-host of the Latent Space Podcast and the co-founder of this very summit. Please join me in welcoming to the stage Swyx. [upbeat music]