# Agents for Everything Else — swyx

Shawn "swyx" Wang

AI Engineer Europe 2026

[Talk page](https://ai.engineer/talks/zepu8Kk6FBQ-agents-everything-else-swyx) · [Video](https://www.youtube.com/watch?v=zepu8Kk6FBQ)

## Agents for Everything Else

Running an AI conference with agents changes more than website development: it shortens design feedback loops, moves schedule management into code, and reshapes who uses software.

### How does a tiny team run an AI conference?

[0:51](https://www.youtube.com/watch?v=zepu8Kk6FBQ&t=51s)

How much of running an AI conference can the organizers hand to AI? At AI Engineer’s first London event, Shawn “swyx” Wang approaches that question through the work of his own team. The opportunity extends beyond writing software: it includes the coordination, setup, and routine knowledge work that determine how much a small organization can accomplish.

His successive AI Engineer keynotes trace the path toward that question. The first, three years earlier, concerned productivity gains from using AI. The next encouraged greater use as costs fell. Swyx describes a roughly 100-fold decline in AI costs every 12–18 months as his working heuristic, rather than a guaranteed rate for every workload.

The third theme was **tiny teams**: teams whose revenue, measured in millions, exceeds their employee count. The accompanying [Tiny Teams Playbook](https://www.latent.space/p/tiny), developed around a World’s Fair track, uses the more specific measure of annual recurring revenue. The attraction is organizational, not just entrepreneurial. A large company can contain tiny teams; the idea does not require finding a lone founder who builds a billion-dollar business.

![Two article excerpts, including “Vibe Coding vs Tiny Teams,” with red boxes highlighting “the rise of Tiny Teams” and “teams with more m in ARR than employees.”](https://ai.engineer/api/talk-essay-media/05961ba4db748bafaa82f10f7ea4f5e31e01064d92ec0d6b2eff7f5ba2a23c8a)

Latent Space’s tiny teams discussion, with its definition highlighted.

Swyx reports that AI Engineer, led by Ben Li and himself, has nine full-time people and runs a business exceeding $9 million. He describes that figure as business size, not ARR. The workflow changes since the organization began two years earlier become his practical example of adopting AI wholeheartedly—complete with physical “AGI pills” given to attendees. His joke is that everyone can now prescribe one to a coworker who remains unconvinced.

### The website begins with a setup problem

[3:07](https://www.youtube.com/watch?v=zepu8Kk6FBQ&t=187s)

The original conference stack was stable and largely non-AI: Figma, React, Supabase, Tito, Google Sheets, and Sessionize. Joining Cognition gave swyx access to coding agents at work, and free access helped prompt serious experimentation. He added [Devin](https://docs.devin.ai/integrations/slack) to the company’s Slack and began showing colleagues how to request changes to the website.

Then the designer brought him a Figma page. The expected implementation cycle was measured in weeks—perhaps one, two, or four. Swyx brought Devin into the conversation, but first faced a prerequisite: connecting it to Figma. He delegated that setup to [Claude Cowork](https://www.anthropic.com/news/introducing-anthropic-labs). Before the coding agent could do the visible work, another agent helped remove the work blocking it.

**An agent’s value includes the prerequisites it removes.** Dependency installation and broken Python environments are concrete examples of this “yak shaving”: each intended action exposes another task that must happen first. Counting only autonomous task completion misses both the depth of that prerequisite chain and the benefit of handling work in parallel.

Once the connection was in place, swyx reports that Devin produced a functioning website matching the Figma design pixel for pixel. He does not give an elapsed implementation time. Seeing the result himself changed his confidence in the tool, and the designer was happy with it. Swyx identifies the result as the conference website live at the time of the talk.

### A shorter feedback loop changes what people attempt

[5:17](https://www.youtube.com/watch?v=zepu8Kk6FBQ&t=317s)

The initial success turned into ongoing collaboration. Swyx points to 207 replies in one Slack thread as evidence of growing use. He would start work before going to bed; the designer, based in Indonesia, would wake up and continue with Devin. The designer began sending red-line annotations, communicating changes visually in the same way he might with a human developer. Swyx compares this with Steve Ruiz’s tldraw examples, but says he had neither taught the designer the technique nor supplied an instruction manual. For a mostly nontechnical team, that independent adoption mattered.

The team also began making things it would normally leave undone. Swyx added a website Easter egg that appears when a visitor on an ultrawide display moves the pointer over the highlights. Then he fed Devin a viral tweet whose design aesthetic he liked, simply to see what it would produce. That experiment grew into a thread with 127 replies. He had initially wanted to avoid spending the designer’s time on a throwaway idea, but the designer joined anyway.

The reason was enjoyment. The wait between an idea and a visible result had largely disappeared: the designer no longer had to block on swyx or a contract developer to try something. Animations, polish, and playful experiments became work the team could initiate directly. **The productivity gain showed up in human initiative, not just generated code.** This is the shift swyx calls agents for everything else.

![An “Agents for Designers” slide with ticket graphics and a Slack thread; three red outlines emphasize requests about visual effects.](https://ai.engineer/api/talk-essay-media/e3b290625edfb7085102b1c4c0048488cb41eb8eaa743a1bea49a54cdfa14cc5)

Design requests and agent replies in an “Agents for Designers” Slack thread.

Figma-to-website and tweet-to-website were useful successes in their own right. Together, they also changed the next question from whether the agent could help to where else the team could use it.

### Make conference data something the agent can maintain

[7:57](https://www.youtube.com/watch?v=zepu8Kk6FBQ&t=477s)

A conference is also a large data-management problem. Swyx describes coordinating 130 speakers, a couple dozen sponsors, and attendees with varied needs. Much of that work resembles maintaining a content management system: collect information, reconcile changes, and keep the published schedule current. The team had experimented with Sanity, but swyx found a more useful arrangement by supplying spreadsheets to Devin and moving the source of truth into code.

With the schedule represented in code, a coding agent could maintain it. Swyx says Devin managed the entire schedule, while emphasizing that another coding agent could fill the same role. His example is a speaker-change email from Marta. Instead of translating the email into a manually edited CMS entry, he could forward it—or paste a screenshot—and ask Devin to handle the change. The incoming message supplied the task context; the agent worked against the schedule’s existing source of truth.

![A Slack speaker-channel message above three schedule screenshots, with a red outline around instructions for requesting speaker changes through Devin.](https://ai.engineer/api/talk-essay-media/e0666b4cd220e32ee44940d4b3e0013979c00145e64ea4573f33dd301232e98a)

“Agents for CMS” pairs speaker-change instructions with schedule screenshots.

Swyx describes a nine-person team managing a conference of roughly 1,000 people. That is his account of operational scale, not an audited attendance count. He then projects handling 6,000 people in San Francisco that summer without increasing the team size. His confidence depends on sufficient onboarding and established workflows: the throughput comes after the agent has been integrated into how the organization works.

The same approach extends to **ETL**, the extraction, transformation, and loading of data between systems. An external vendor holds information missing from the central source of truth. The task is to obtain the vendor’s API key, synchronize its data, and reconcile the information into that shared source. This is routine integration work, but it is also part of the conference operation that an agent can take on.

### Finding a lobster in London

[9:33](https://www.youtube.com/watch?v=zepu8Kk6FBQ&t=573s)

The next example leaves software maintenance entirely. A viral post showing a claw beside the Wall Street bull inspired swyx to put a similar prop in front of the conference. He asked Devin: “Where can I get a lobster in London?” The agent returned phone numbers, email addresses, and websites. Swyx clicked through the options, considered them, and requested more research. He credits that process with sourcing the lobster used at the conference; the described workflow includes his evaluation of the results, rather than an unattended purchase.

For this kind of task, his requirements are simple: web access and a sufficiently capable model. Whether the assistant is called [OpenClaw](https://openclaw.ai/), [NanoClaw](https://github.com/nanocoai/nanoclaw), or something else matters less than whether it can do knowledge work he would otherwise assign to an executive assistant or junior employee. A coding agent becomes an on-demand worker for research and coordination as well as implementation.

### From planning notes to speaker research

[10:44](https://www.youtube.com/watch?v=zepu8Kk6FBQ&t=644s)

Swyx broadens the examples beyond Devin, disclosing that he now advises Cognition rather than working there in the capacity described earlier. He has begun exploring [Town](https://www.town.com/) as coding agents expand into general knowledge work. He connects this direction with purpose-built knowledge tools, Andrej Karpathy’s discussions of wikis, and what he describes as OpenClaw’s adoption of that approach. He expects this expansion to be one of the leading trends of 2026.

For World’s Fair planning, he starts with Apple Notes containing prospective speakers and tracks. He supplies those notes to Town and gets a formatted Notion document containing research on the speakers he is considering approaching and curating. The names on the displayed notes are intentionally small; the useful part of the example is the transformation from a personal planning list into researched material that supports the next decision.

### The people who inherit a replacement must trust it

[11:43](https://www.youtube.com/watch?v=zepu8Kk6FBQ&t=703s)

Success with individual tasks can lead to a larger ambition: replacing an entire SaaS product with something built internally. Swyx shows himself debating an unnamed replacement with employees. He jokingly calls his stronger enthusiasm “psychosis,” but takes the disagreement seriously. A manager cannot dismiss the objections of the people who will have to operate the result.

Those concerns are valid because employees bear the consequences when the replacement fails—and swyx explicitly acknowledges that the team gets things wrong. His method is to identify the three most important concerns and systematically reduce them. At the time of the talk, that evaluation is still underway. The example is an approach to earning confidence in a replacement, not a completed story of eliminating a SaaS dependency.

### Build interfaces for the user that is arriving

[12:44](https://www.youtube.com/watch?v=zepu8Kk6FBQ&t=764s)

These internal changes also suggest a product-design problem: the entity using a service may increasingly be an agent. Swyx attributes to Malte’s opening keynote a claim that 60% of Vercel’s user base consists of bots or agents. The denominator and measurement period are not established here, so that figure should not be treated as a measured share of customers. It motivates swyx’s broader emphasis on APIs, CLIs, and MCP interfaces as ways for agents to use software.

There is a related change in where interfaces appear. Swyx points to MCP Apps speakers Ido and Liad, also referencing their ETN appearance, as advocates for shipping UI into someone else’s application. [MCP Apps](https://blog.modelcontextprotocol.io/posts/2026-01-26-mcp-apps/) supplies a concrete setting for that idea: users can encounter a service’s interface within an MCP client. A product’s own dashboard no longer has to be the sole destination for using it.

**Agent experience means designing for this changing primary user.** Swyx’s own Figma workflow makes the shift tangible: he cares less about navigating the dashboard himself when he can give the task to Claude Cowork and hope it completes the work. His closing invitation is to bring that experimentation into everyday work, including the tasks surrounding software development. The AGI-pill joke returns as an invitation to bring hesitant colleagues along.

## Resources

- [The Tiny Teams Playbook](https://www.latent.space/p/tiny) — Swyx's synthesis of hiring, culture, operations, and technical practices from the World's Fair Tiny Teams track.
- [Use Devin in Slack](https://docs.devin.ai/integrations/slack) — Current setup and collaboration instructions for starting Devin sessions and sharing attachments through Slack.
- [Devin MCP Use Cases](https://docs.devin.ai/use-cases/tutorials/mcp-use-cases) — Examples of connecting Devin to external tools, including working with Figma designs.
- [Town](https://www.town.com/) — An AI assistant for delegated work that maintains context about how its user works.
- [AI Index 2025: Inference Cost Trends](https://hai.stanford.edu/assets/files/hai_ai-index-report-2025_chapter1_final.pdf) — Benchmark-specific evidence on falling inference prices, including GPT-3.5-level performance on MMLU.
- [MCP Apps](https://blog.modelcontextprotocol.io/posts/2026-01-26-mcp-apps/) — The official announcement explaining how interactive application interfaces appear inside MCP clients.
- [NanoClaw](https://github.com/nanocoai/nanoclaw) — A customizable personal assistant whose agents run in separate Linux containers.
- [OpenClaw](https://openclaw.ai/) — The personal AI assistant project mentioned as another way to delegate everyday knowledge work.

## Slides

### AI Engineer Europe

[0:00](https://www.youtube.com/watch?v=zepu8Kk6FBQ&t=0s) [AI Engineer Europe logo in white on a black background.](https://assets.ai.engineer/sites/aiecode2025/sha256/aab59c7da120f8504a96c9b8622254406bbff88706e2388a912d6dcc744c1118.webp)

AI Engineer
EUROPE

### Presenting sponsor: Google DeepMind

Google DeepMind is the presenting sponsor.

[0:05](https://www.youtube.com/watch?v=zepu8Kk6FBQ&t=5s) [Google DeepMind presenting sponsor card on a black background.](https://assets.ai.engineer/sites/aiecode2025/sha256/22e8cb940ef752894fb0f47ce4cfbb7a04b212fd252bdf1347f3f5879fa393f3.webp)

PRESENTING SPONSOR
Google DeepMind

### Platinum sponsors

Braintrust, WorkOS, and OpenAI are the platinum sponsors.

[0:10](https://www.youtube.com/watch?v=zepu8Kk6FBQ&t=10s) [Platinum sponsors Braintrust, WorkOS and OpenAI on a black background.](https://assets.ai.engineer/sites/aiecode2025/sha256/7f3ed063d777a261c891941314e6d2625c0c8750b434b0963aa0b8bb198218fc.webp)

PLATINUM SPONSORS

Braintrust
WorkOS
OpenAI

### Building AI Engineer with AI

swyx introduces how the team builds AI Engineer with AI, describing it as perhaps his biggest revelation.

[0:47](https://www.youtube.com/watch?v=zepu8Kk6FBQ&t=47s) [Opening slide for swyx’s AI Engineer Europe 2026 address.](https://assets.ai.engineer/sites/aiecode2025/sha256/485f45e7982d37969ec18f6e650b5990b1ed16bc1921591353794dc0ad5a03c9.webp)

Building AI Engineer with AI

AI Engineer Europe 2026
Presented by
swyx

Footer:
AI Engineer Europe 2026
Building AI Engineer with AI
swyx

Broadcast text:
OPENING ADDRESS
SWYX / Co-founder, Editor
AI Engineer
LATENT SPACE

AI Engineer EUROPE
PRESENTED BY
Google DeepMind

### The 1,000x AI Engineer

Swyx revisits his first keynote’s model: improvements from AI-enhanced engineers, AI product engineers and AI engineer agents stack from 10x to 1,000x.

[1:17](https://www.youtube.com/watch?v=zepu8Kk6FBQ&t=77s) [A video screenshot presents three stackable improvements: AI-enhanced engineers, AI product engineers and AI engineer agents.](https://assets.ai.engineer/sites/aiecode2025/sha256/eeb3c02762b4ee7ec73080e3752b235de283f9a05140fda1fa6e098c6f00b42e.webp)

The 1000x AI Engineer = Stackable 10 x 10 x 10 improvements

10x AI Enhanced Engineer
• Copilot, Ghostwriter, Codium
• Build your own dev tools
• Use Knowledge Work AI tools

10x AI Product Engineer
• RAG/LLM frameworks
• Models/Infra/Finetuning
• AI UX & Growth Hacking

10x AI Engineer Agent
• Writes faster than you
• While you sleep
• In parallel, on demand

Illustration text: [illegible]
AIE
FIXIE
supabase

AI Engineer Europe 2026
Building AI Engineer with AI
swyx

Interface text:
Search
The 1,000x AI Engineer: Swyx
AI Engineer
392K subscribers
488
Share
Ask
Save

### Cost of Intelligence

Swyx says AI costs have fallen roughly 100-fold every 12–18 months and thinks the trend is continuing. The slide pairs the Mixtral inference price war with a comparison of closed-source and open-weight model performance.

[1:34](https://www.youtube.com/watch?v=zepu8Kk6FBQ&t=94s) [An article and chart about falling AI inference prices appear beside a comparison of closed-source and open-weight models.](https://assets.ai.engineer/sites/aiecode2025/sha256/5a3f843344c1ae8c5d5fc9ffcfbcdc83f73d1c941a0bda18a8473c3c2b5a0fe3.webp)

Cost of Intelligence

Latent Space

Mixtral sparks a GPU/Inference Race to the Bottom

The GPU/Inference War had been brewing all year, with Together, Fireworks, Perplexity, and others raising hundreds of millions to build up infrastructure businesses - and the release of Mixtral was the spark that set everything aflame. We've seen all of the major inference players slashing prices, undercutting Mistral’s own la plateforme pricing by as much as 70% overnight.

Cost of Different AI Models (in $ per Million Output Tokens)
>222x cheaper in the last 2 years
Mixtral Price War of Dec 2023
RIP Margins
Model/provider labels: [illegible] (2020); text-davinci-003 (Nov 2022); gpt-3.5-turbo (Mar 2023); Mixtral on Mistral; Fireworks AI; Together AI; Perplexity AI; Anyscale Endpoints; Abacus AI; DeepInfra
We’ve now seen a 99.55% drop in the price of GPT3/3.5-level artificial intelligence

Maxime Labonne
@maximelabonne

Closed-source vs. Open-weight LLMs

The gap between the best-performing closed-source and open-source models is also narrowing in terms of Arena ELO (chat.lmsys.org/?leaderboard).

We now face a 6 to 10-month lag, rather than years as was the case when GPT-4 was released.

Let’s see how far Llama 3 takes us!

Closed-source vs. Open-weight models (Arena ELO)
Legend: Closed-source models; Open-weight models
Y-axis: Arena ELO
X-axis: Release Date
Chart labels and ticks: [illegible]

Interface text:
2:19 AM · Apr 15, 2024 · 151.8K Views

### Vibe Coding vs Tiny Teams

Swyx defines Tiny Teams as small, productive teams augmented by coding agents, with an aspirational measure: more millions in ARR than employees.

[1:48](https://www.youtube.com/watch?v=zepu8Kk6FBQ&t=108s) [Two article excerpts highlight the rise of Tiny Teams and define them by revenue relative to employee count.](https://assets.ai.engineer/sites/aiecode2025/sha256/9c6303e1307830585d1e4123e7a4bb9f89a90028399364b6a01097ea45faab76.webp)

Latent Space

SWE Agents Too Cheap To Meter, The Token Data War, and the rise of Tiny Teams

Cheap SWE Agents are enabling teams with more millions in ARR than employees. Reflections after the OpenAI Codex and Google Jules Launches, Cognition DeepWiki, and LMArena's $100m raise.

MAY 24, 2025

btw Gamma, Bolt, Oleve and more are presenting on the new Tiny Teams track, and leads of both OAI Codex and Google’s Jules will be presenting at AI Engineer World’s Fair! You know the drill by now - the AIE expo and all 3 conf hotels are now sold out! Join the Hackathon and fill out the State of AI Engineering Survey!

The recent launches of the Codex and Jules Devin-like autonomous coding agents came with some interesting pricing decisions: they come at no extra cost¹. You no longer have a reason NOT to send off every thought to the agents (even more than once) even if you intend to eventually code it yourself.

Let me repeat: YOU CAN FIRE OFF A CODEX REQUEST EVERY 60 SECONDS AND THEY DON’T CHARGE YOU EXTRA. YOU SHOULD USE THIS!!!

This comes after a 3 year trend of exploring the SWE Agent pricing spectrum:

Coding Agents with ~fixed pricing plans*
Y-axis label: $500/month
DEVIN
Claude Max
lovable

Latent Space

Vibe Coding vs Tiny Teams

A few people have asked my opinion on the vibecoding trend. At almost every SF house party, a technical engineer or three will sheepishly admit that they’ve been vibe coding, like it’s some sort of vice. We even went ahead and put in a Vibe Coding category in the AIE CFP. But ultimately we nixed it in favor of “Tiny Teams”: the rising trend of small but extremely productive teams augmented by SWE Agents (not necessarily making AI products). The simplest (but hardest) statement of this is “teams with more m in ARR than employees”, which is rarefied air but already reached by @levelsio⁶, Cognition, Bolt, Gumloop and Gamma.

Tiny Teams are the scaled form of first type of AI Engineer predicted in The 1000x AI Engineer, and is the more realistic near term version of the “one person unicorn” that every thoughtleader is fantasizing about. Tiny Teams are easily profitable - most tech businesses are 60%-90% margin and most cash compensation should be <$250k - and therefore are also likely to lead to Pegasus and other outcomes.

The way I put it to a friend recently is that “vibe coding” is a slop attractor, while “Tiny Teams” focuses on productivity. One might look askance at Rick Rubin’s new book on Vibe Coding as one such instantiation, whereas it’s pretty hard to deny that a Tiny Team (in the way we have defined it) is an unambiguously Good Thing and a direct consequence of declining end-user costs in SWE Agents.

Interface text: 36; 3; 6; Share

### AI Engineer event growth

AI Engineer reports attendance rising from 500+ in 2023 to 5,000+ in 2025, while views rose from 320K to 9.1 million. The 2026 column lists 9,000+ attendees and 2.4 million views as of April 9.

[2:22](https://www.youtube.com/watch?v=zepu8Kk6FBQ&t=142s) [A conference video shows attendance and view figures for 2023 through 2026.](https://assets.ai.engineer/sites/aiecode2025/sha256/fb8866d496ac77deadc31e1ca9133286ce99cd9db575b9988017b27d2b67f5b9.webp)

900% Attendees & 2,744% Views*

2023
500+ Attendees
320K VIEWS
AI Engineer SUMMIT

2024
3,000+ Attendees
1.1 MILLION VIEWS
AI Engineer World’s Fair

2025
5,000+ Attendees
9.1 MILLION VIEWS
AI Engineer SUMMIT
AI Engineer World’s Fair
AI Engineer Code Summit

2026
9,000+ Attendees
2.4M VIEWS (A/O APRIL 9)
AI Engineer EUROPE
AI Engineer World’s Fair
AI Engineer NEW YORK
AI Engineer Code Summit

*From 2023 to 2025

WELCOME TO AI ENGINEER EUROPE
LIA MCBRIDE / General Manager
AI Engineer

AI Engineer EUROPE
PRESENTED BY
Google DeepMind

Interface text:
1.00
AIE Europe Day 1: Keynotes & OpenClaw/Personal Agents ft Google Deepmind, OpenAI, Vercel, & more
AI Engineer
392K subscribers
324
Share
Ask
Save

AI Engineer Europe 2026
Building AI Engineer with AI
swyx

### The AI Engineer team

A team photo sits beside AI Engineer’s channel member list. swyx says nine full-time people run a business of more than $9 million.

[2:31](https://www.youtube.com/watch?v=zepu8Kk6FBQ&t=151s) [A team photo beside a workplace channel member list.](https://assets.ai.engineer/sites/aiecode2025/sha256/75b6ffe674eaf911f52a1f8060c38d0795a8750fd5951d48693c1b92c1fb9a07.webp)

# all-internal-aie

ben — Channel Manager
Eliana — Eliana Santos
Sales Director
Kelsey Mohland — Kelsey Mohland
Head of Ops
Kyle Hollinger — Kyle Hollinger
Production, Programming & Operations Manager
Lia McBride — Lia McBride
GM
Natalie Morrow-Newman — Natalie Morrow-Newman
Producer
Sherine Whittle
Producer
swyx (AIE) (you) — Shawn Wang (AIE)
cofounder & curator
Tan — Tan
Trish — Trish
Vincent (Design) — Vincent (Design)

AI Engineer Europe 2026
Building AI Engineer with AI
swyx

Interface text:
All new posts
Huddle
About
Members 11
Tabs
Integrations
Settings
Find members
Everyone

### Taking the AGI Pills

swyx introduces his AGI Pills giveaway: bottles filled with sweets, joking that attendees can prescribe them to coworkers who need more AGI enthusiasm.

[2:42](https://www.youtube.com/watch?v=zepu8Kk6FBQ&t=162s) [A social post shows a novelty AGI Pills giveaway beside the slide title.](https://assets.ai.engineer/sites/aiecode2025/sha256/0f565a669f63b4a0a4e62e2c02da951bfa4e82928a9706c3beac59aa762b1125.webp)

Taking the AGI Pills

Luke Knight @lukeknight · 2h
Please save me a bottle of agi pills @swyx

STEP 1  Grab Your Bottle
STEP 2  Affix your label
STEP 3  Fill it up with sweets!

Intelligence. Compressed.
AGI Pills
Warning: [illegible]

AI Engineer @aiDotEngineer · 3h
💊 Have you got your @aiDotEngineer AGI Pills prescription from DR @swyx yet?

AI Engineer Europe 2026
Building AI Engineer with AI
swyx

Interface text:
0:37
1
5
2K

### Our Stack (up to 2025)

Up to 2025, AI Engineer used a stable, non-AI stack across marketing, ticketing, sponsors and speakers. After joining Cognition, swyx started using coding agents seriously at work.

[3:06](https://www.youtube.com/watch?v=zepu8Kk6FBQ&t=186s) [A list of tools used for marketing, ticketing, sponsors and speakers.](https://assets.ai.engineer/sites/aiecode2025/sha256/91f896dd7797e262dc30f8928f55ea3849924d988433e0ffe9d3d9afc4066ee0.webp)

Our Stack (up to 2025)

- Marketing: Figma + React + Vercel + Supabase
- Ticketing: Ti.to + Validar
- Sponsors: Figma + Google Sheets
- Speakers: Sessionize

AI Engineer Europe 2026
Building AI Engineer with AI
swyx

### Joining Cognition

After joining Cognition, swyx began using coding agents seriously at work, mostly because they were free. He added them to company Slack and showed colleagues how to use them on the company website.

[3:21](https://www.youtube.com/watch?v=zepu8Kk6FBQ&t=201s) [An article about joining Cognition appears beside related social posts.](https://assets.ai.engineer/sites/aiecode2025/sha256/509b5d5819dd741c6461b0685ddc940d36a8ec1774990f58162d2d129247b8b4.webp)

Cognition: The Devin is in the Details
swyx · 2025-09-08

thanks to Turing Post and AI Breakdown for covering this! Twitter too. Thanks Niko for picking up on the Agent Lab thesis - it has since been updated on Latent Space!

Not to bury the lede: I'll be joining Cognition, makers of the Devin AI software engineer and now owners of the Windsurf agentic IDE. Most of Smol AI will be joining Cognition as part of the deal, while I will continue to work on AI Engineer and Latent Space independently of Cognition (As part of restructuring, I'm delighted that Lia is stepping up to be General Manager of AI Engineer).

Note: I've been fairly public that I've advised companies before, most recently Fireworks, and while I will ferociously defend the neutrality and independence of AIE and LS, obviously this new Cognition deal is quite a bit bigger than others, so I'm sharing some thoughts on occasion of this fundraise, both for my future self and others scrutinizing this choice.

When I first tried Devin in April 2024 I got a lot of shit for leaning positive, when all people wanted to talk about was the Upwork video. But I gave them a lot of grief in private, which is, btw, how you're actually supposed to conduct yourself in polite company. Fast forward and Cognition is announcing that they have gone from 0 to their $10B Series C in 2 years today, with me effectively packing up and joining them [1].

I wanted to explain why I am basically putting my Smol “chips” into Cognition despite such a historically outsized runup in valuation:
1. Nontechnical thesis: Short Code, Long AGI timelines
2. Agents thesis: The rise of Agent Labs
3. Engineer thesis: Owning the Sync/Async spectrum
4. Business model: Usage Based for Enterprise, Cheap Seats for Individuals
5. Team: Cracked Engineers + Ramped GTM

swyx 🇬🇧 @aidotengineer · @swyx · Sep 8, 2025
ok life update: i'll be joining @Cognition!
• Cog just went from 0 to $10b in 2 years
• Net burn $20m in company history
• Avg successful Devin impl sees >5x growth, $1.5m/yr custo[illegible]
>10x in 8 months (not typo)
• Windsurf x Cog cross sell going great, looking fwd

Embedded image: Cognition: The Devin is in the Details. [illegible]

Cognition @cognition · Sep 8, 2025
We've raised over $400M at a $10.2B post-money valuatio[illegible]
the frontier of AI coding agents.

The round was led by Founders Fund with other existing in[illegible]
including Lux, 8VC, Neo, Elad Gil, Definition Capital, and S[illegible]

Interface text: Home · Ideas · About · Subscribe · Show more

### Agents for Designers: Design Variations

A designer presents two website concepts in Figma: a condensed dark design and a more visual marketing design. swyx says the designer expected turning the design into reality to take one to four weeks.

[3:46](https://www.youtube.com/watch?v=zepu8Kk6FBQ&t=226s) [A Slack thread describes two website design variations and invites feedback in Figma.](https://assets.ai.engineer/sites/aiecode2025/sha256/815ad205f0cf8da2dc0b21800b1c23dd5c8df30f06f5b2eb239e995fdb8b18be.webp)

Agents for Designers

[Inactive Account] Jan 28th at 7:27 AM
Hi @swyx (AIE) @ben, here are 2 variations for WF'26 website design.

The first one is more condensed, text-heavy, and dark-mode. You can check it out here

The second one is more marketing-centric, with more balance between images & text with a combination of light and dark background. More like our current WF'25 website but more polished. You can check it out here

Please check it out when you have a chance. Figma link for holistic view:
https://www.figma.com/design/[illegible]/World-s-Fair-2026?node-id=140-3540&t=[illegible]

Figma
World's Fair 2026
Created with Figma

Figma
World's Fair 2026
Created with Figma

162 replies

[Inactive Account] Jan 28th at 7:56 AM
We could add a zoom effect to the second option when user scroll from the hero section to below, the video frame starts growing and becoming full screen similar to the one from realfood.gov

Also let me know if we should try something more unique and out of the box (edited)

ben VIP Jan 30th at 2:59 PM
fyi shawn i'm commenting/working with vincent in figma on this, i've invited you if you want to see/add your own feedback/ideas

Interface text:
Thread
AI Engineer Europe 2026
Building AI Engineer with AI
swyx

### Agents for Designers: Design Feedback

After shipping a simple version of design 2, swyx asks the designer to annotate fixes, especially for mobile. He is unsure how Devin can access the Figma feedback and suspects it needs Figma MCP.

[3:59](https://www.youtube.com/watch?v=zepu8Kk6FBQ&t=239s) [A Slack thread requests design feedback and discusses how Devin might access Figma.](https://assets.ai.engineer/sites/aiecode2025/sha256/1bbe27a2070661f59c8addc31b0a9deaae57240eb89def1b11c3fcb11152b9de.webp)

Agents for Designers

swyx (AIE) — Feb 2nd at 6:07 PM
@[Inactive Account] a simple version of design 2 has been shipped live. https://www.ai.engineer/worldsfair could you take a close look at especially the mobile view, and document everything you would fix? maybe in figma, screenshot what we have live now on website, and what you would change. do as many little arrows and comments as possible to help me use Devin to fix it up

[Inactive Account] — Feb 3rd at 12:20 AM
Wow, that was fast. Sure thing, I'll review the website

@swyx (AIE) can devin access figma file? What's the best way to provide the feedback to Devin?

swyx (AIE) — Feb 3rd at 8:05 AM
@[Inactive Account] honestly, not sure. this is mostly due to my lack of experience doing this. my guess is that i need to set up the Figma MCP but also it will probably be bad at it

(aka have to have me look at it and translate to commands for devin)

AI Engineer Europe 2026
Building AI Engineer with AI
swyx

Interface text:
Thread
Also sent to the channel

### Setting Up Figma MCP with Cowork

swyx asks Cowork to read the Figma MCP guide and Devin’s design implementation guide, then help set up Figma MCP in Devin.

[4:03](https://www.youtube.com/watch?v=zepu8Kk6FBQ&t=243s) [A Cowork prompt asks for help setting up Figma MCP in Devin using documentation.](https://assets.ai.engineer/sites/aiecode2025/sha256/7ae6d34172e20b70906ebdd6207f3e2f8138c5a2b3e42c4427c83ec8840e9e6b.webp)

Let's knock something off your list

Cowork is in research preview. Learn how to use it safely.

hey cowork, read https://help.figma.com/hc/en-us/articles/32132100833559-Guide-to-the-Figma-MCP-server and https://docs.devin.ai/use-cases/gallery/figma-design-implementation and help me get the figma MCP setup in Devin

Interface text:
Chat
Cowork
Code
Work in a project
Opus 4.6
Let's go

### Yak Shaving With and Without Agents

The diagram contrasts a task stalled by nested dependencies with agents handling those dependencies in parallel, freeing the person to complete one task and move to another within a focus window.

[4:22](https://www.youtube.com/watch?v=zepu8Kk6FBQ&t=262s) [A diagram compares sequential work interrupted by dependencies with agents handling those dependencies alongside tasks.](https://assets.ai.engineer/sites/aiecode2025/sha256/a3ce1769848eda720878126ba508f815b0d2f60018d2df180b4af4d78a6ce6e9.webp)

Without Agents:
Everything must be done in sequence, by yourself.
Yak shave = Things get dropped!

With Agents:
- agents can handle yak shaves
- agents complete shaves faster
- you can parallelize/ctx switch

Without Agents diagram:
max time allocated in focus window
Task 1
Incomplete
Yak shave 1
resume
Yak shave 2

With Agents diagram:
max time allocated in focus window
Task 1
complete!
agents handle yak shaves
Task 2
agents handle yak shaves

### Agents for Designers: Implemented Feedback

After connecting Devin to Figma, swyx says the team got a working website and the designer was happy. Devin reports 12 design fixes and a PR with CI passed.

[4:46](https://www.youtube.com/watch?v=zepu8Kk6FBQ&t=286s) [A Slack message reports website design fixes made from Figma feedback, with desktop and mobile preview images.](https://assets.ai.engineer/sites/aiecode2025/sha256/401d2de25b1d7ca5fc429e0ca10af925f4718858083b6faef2624c434d8bc0ee.webp)

Agents for Designers

Vincent (Design) — Feb 4th at 8:47 AM
@Devin did you see this comment "Update speaker logos from this file https://drive.google.com/file/d/[illegible]/view?usp=drive_link" on World's Fair 2026 figma file?

Devin APP — Feb 4th at 8:47 AM
PR created and CI passed: https://github.com/aiDotEngineer/aiecode2025/pull/93

Vercel Preview: https://[illegible].vercel.app/worldsfair

12 design fixes implemented from Vincent's Figma feedback:

Desktop:
• JetBrains Mono font applied to WF2026 page
• Logo height increased to 64px
• "BUY TICKETS" button now white by default, transparent on hover

Footer:
• Section labels now visible in white
• Social icons now white
• Newsletter input has dark background (#1A1A1A)

Mobile:
• Hero text center-aligned
• Reduced spacing by ~50%
• Speaker logos marquee animation
• Speakers section text center-aligned

Other:
• Ticket iframe has 4px border radius

You can review the changes in the Devin session at https://app.devin.ai/sessions/[illegible] or leave comments directly on the PR.

Preview images:
World's Fair
THE ONLY CONFERENCE AI ENGINEERS MUST ATTEND IN 2026
Engineering the future of AI
6,000+   300   100   30+
HEAR FROM LEADERS AT THE WORLD'S TOP AI AND TECH COMPANIES.
ANTHROPIC
[illegible]

Interface text:
4 files
Download all
AI Engineer Europe 2026
swyx

### Agents for Designers

After an initial success with Devin, the team kept using it. The design discussion grew to 207 replies.

[5:11](https://www.youtube.com/watch?v=zepu8Kk6FBQ&t=311s) [A design discussion compares minimal and scrollytelling approaches for the Europe event website.](https://assets.ai.engineer/sites/aiecode2025/sha256/4b73c2e1cfb172398c681bc55fadeed80da5c0576b6905a77efdc508929a0f4f.webp)

Agents for Designers

[Inactive Account] 8:38 AM
@swyx (AIE) @ben the first version is a minimalistic approach, inspired by AIE Code last year, where the content is minimal, straight to the point. Maybe we could add simple animation between elements if we choose this approach.

Figma: https://www.figma.com/design/[illegible]/AIE-Europe-2026?node-id=93-241&t=[illegible] (edited)
image.png

Europe’s first official AI Engineer event
Speakers
[illegible]

207 replies  Last reply 2 months ago

[Inactive Account] 8:40 AM
The second version is a scrollytelling approach. Here's the video sample of how the first part might look
@swyx (AIE) @ben

Figma: https://www.figma.com/design/[illegible]/AIE-Europe-2026?node-id=105-144&t=[illegible]
aie-europe-hero-sample.mp4

Interface text:
Monday, January 19th
World’s Fair
SAN FRANCISCO
JUNE 30–JULY 2, 2026
Engineering the future of AI
AI Engineer Europe 2026
Building AI Engineer with AI
swyx

[5:21](https://www.youtube.com/watch?v=zepu8Kk6FBQ&t=321s) [The design discussion highlights a thread with 207 replies.](https://assets.ai.engineer/sites/aiecode2025/sha256/3ab8536ed0f8350680a8d2d3e45ddbaaec94b56c9c4d8824254e3239695537c6.webp)

Agents for Design

[Inactive Account] 8:38 AM
@swyx (AIE) @ben the first version is a minimalistic approach, inspired by AIE Code last year, where the content is minimal, straight to the point. Maybe we could add simple animation between elements if we choose this approach.

Figma: https://www.figma.com/design/[illegible]/AIE-Europe-2026?node-id=93-241&t=[illegible] (edited)
image.png

Europe’s first official AI Engineer event
Speakers
[illegible]

207 replies  Last reply 2 months ago

[Inactive Account] 8:40 AM
The second version is a scrollytelling approach. Here's the video sample of how the first part might look
@swyx (AIE) @ben

Figma: https://www.figma.com/design/[illegible]/AIE-Europe-2026?node-id=105-144&t=[illegible]
aie-europe-hero-sample.mp4

Interface text:
Monday, January 19th
World’s Fair
SAN FRANCISCO
JUNE 30–JULY 2, 2026
Engineering the future of AI
AI Engineer Europe 2026
Building AI Engineer with AI
swyx

### Agents for Designers: Feedback

Vincent says Devin makes it easy to see his designs come to life and removes back-and-forth with an engineer. swyx describes the faster feedback cycle as a reason the designer kept working on the project.

[6:44](https://www.youtube.com/watch?v=zepu8Kk6FBQ&t=404s) [A designer describes the ease of seeing a design come to life through an agent.](https://assets.ai.engineer/sites/aiecode2025/sha256/8b76d3aea2bd7493e8deedea2408e9f2ae659ea56d4b28a34210a8e6ee739bda.webp)

Agents for Designers

[illegible] comments on the PR for further tweaks. Would you like me to record a visual walkthrough?

Vincent (Design)  Mar 3rd at 4:10 AM
Looks great @Devin another adjustment:
• Increase spacing on confirmed speakers header see attachment
• Not this small for platinum sponsors logo, at least 2 times the current size
• Could you add a toggle on the right navbar to toggle between dark mode and light mode? light mode is this version and dark mode is the one we have before we change it to white background and stuff

2 files  Download all
Confirmed Speakers
PLATINUM SPONSORS

Devin  APP  Mar 3rd at 4:10 AM
Working on all 3 — speakers header spacing, bigger platinum logos, and a dark/light mode toggle. The toggle is the most involved since it needs to switch between the current light [illegible]

swyx (AIE)  Mar 3rd at 4:17 AM
why are u so excited by this haha

Vincent (Design)  Mar 3rd at 4:22 AM
lol maybe it's because of how easy and pixel-perfect it is to see your design come to life

I mean usually when I was handing over my design to a real human engineer, there was a lot of back and forth and now it just removes that barrier

swyx (AIE)  Mar 3rd at 4:22 AM
its fun right

Vincent (Design)  Mar 3rd at 4:23 AM
yeah haha

do u think we should add this as a easter egg
put this under a light mode toggle

Devin  APP  Mar 3rd at 4:24 AM
All 3 adjustments done, CI green: aiecode2025#191 (Devin Review)

Interface text:
Thread
AI Engineer Europe 2026
Building AI Engineer with AI
swyx

## Full transcript

[0:00](https://www.youtube.com/watch?v=zepu8Kk6FBQ&t=0s) [upbeat music] The future of work has many paths.

[0:18](https://www.youtube.com/watch?v=zepu8Kk6FBQ&t=18s) Our next presenter will discuss the path that he walked with Devin as he organized this very conference. Please join me in welcoming to the stage the co-founder of AI Engineer Conferences, swyx. [upbeat music]

[0:33](https://www.youtube.com/watch?v=zepu8Kk6FBQ&t=33s) Oh. [audience applauding] Hi, everyone. Uh, I am not the chief AI officer of the UK. Uh, I-- unfortunately, he, he had to leave for a personal reason. Uh, but, you know, you get me.

[0:47](https://www.youtube.com/watch?v=zepu8Kk6FBQ&t=47s) Uh, thanks for staying so long. I hope-- [audience applauding] Is everyone having a good time? Thank you.

[0:51](https://www.youtube.com/watch?v=zepu8Kk6FBQ&t=51s) Uh, it, it's so endearing and, and, and heartwarming to, to hear from you guys. Uh, I'll take you a little bit into how we build AI Engineer with AI, and it's, it's probably the biggest revelation that I've had.

[1:05](https://www.youtube.com/watch?v=zepu8Kk6FBQ&t=65s) Uh, so yeah, we've had, we've had a lot of really warm reception from you guys, and I think it's really great. And, uh, I think this is something that we really try to engineer, and, and hopefully, you know, this is our first event in London.

[1:16](https://www.youtube.com/watch?v=zepu8Kk6FBQ&t=76s) Hopefully, you have us back next year. Um, one, one thing I wanted to, so for those who are newer to us, uh, I do one of these keynotes every single AIE.

[1:25](https://www.youtube.com/watch?v=zepu8Kk6FBQ&t=85s) Um, the, the first, very first one three years ago, I s- talked about the productivity gain that you get from the increased AI eng-- uh, from, from the increased usage of AI.

[1:35](https://www.youtube.com/watch?v=zepu8Kk6FBQ&t=95s) Um, and the second one, we talked about how you should just use more AI because the cost curve of AI is going down roughly a hundred times, uh, per, per every twelve to eighteen months, and I think it's still continuing to, to trend that way.

[1:50](https://www.youtube.com/watch?v=zepu8Kk6FBQ&t=110s) Um, the third year, we started to talk about tiny teams, uh, which, which was basically this definition that I had that teams with more millions in revenue than number of employees.

[2:00](https://www.youtube.com/watch?v=zepu8Kk6FBQ&t=120s) Um, and I even curated an entire track at the World's Fair about this, uh, where we ta-- where we sort of summarized it as the Tiny Teams playbook if you're interested in building that.

[2:09](https://www.youtube.com/watch?v=zepu8Kk6FBQ&t=129s) Uh, the reason I liked this emphasis is because, um, I think people are maybe too egotistical about looking at-- for the one-person billionaire or, or unicorn founder. Um, every company can have a tiny team, whether you're small or large.

[2:23](https://www.youtube.com/watch?v=zepu8Kk6FBQ&t=143s) Um, and I think when I look at how I-- how we run AI Engineer, uh, me being the, the, the leadership of Ben Li and myself, uh, we are also a tiny team.

[2:31](https://www.youtube.com/watch?v=zepu8Kk6FBQ&t=151s) Um, this is us. Uh, it's just, uh, nine full-time people, and, uh, we're running a business that is more than nine million dollars, so we are a tiny team.

[2:43](https://www.youtube.com/watch?v=zepu8Kk6FBQ&t=163s) Um, and I wanted to show you the most significant changes in our workflow, uh, since we started this two years ago. Uh, by the way, this is, uh, our taking the AGI pill moment.

[2:53](https://www.youtube.com/watch?v=zepu8Kk6FBQ&t=173s) Uh, did you guys get the AGI pills?

[2:55](https://www.youtube.com/watch?v=zepu8Kk6FBQ&t=175s) Yeah.

[2:55](https://www.youtube.com/watch?v=zepu8Kk6FBQ&t=175s) Yes. [laughs] I'm very proud of this. This is my brainchild. Uh, if one of your coworkers is not sufficiently AGI pilled, you should prescribe one of this. You are all AGI doctors now.

[3:07](https://www.youtube.com/watch?v=zepu8Kk6FBQ&t=187s) Okay. Our stack was very stable and completely non-AI, which is very ironic for an AI conference. Uh, we do Figma, React, Supabase, Tito, Fig-- uh, Google Sheets, Sessionize. Um, and then I had this funny, weird moment where I joined Cognition, uh, and, and started talking, uh, started using coding agents seriously at work, mostly because they were free.

[3:31](https://www.youtube.com/watch?v=zepu8Kk6FBQ&t=211s) Um, and I started adding it to the company Slack, and then I started doing things with it and showing people, "Hey, here's how you use it to do coding on the, the company website."

[3:41](https://www.youtube.com/watch?v=zepu8Kk6FBQ&t=221s) All well and good, and something strange starts happening. Um, I start introducing-- uh, this is, this is a, a c- a workflow of our contract designer, now full-time, um, uh, showing me a Figma page and asking me to go through it and expecting that we would take a week, two weeks, four weeks to turn it into reality.

[4:00](https://www.youtube.com/watch?v=zepu8Kk6FBQ&t=240s) Um, I just added Devin to it, um, and ul- and ultimately, uh, before I had to add Devin to it, I had to hook up Devin to Figma, and I'm not gonna be doing that shit.

[4:10](https://www.youtube.com/watch?v=zepu8Kk6FBQ&t=250s) So Cowork is doing it for me. You should use Cowork for, [laughs] uh, for doing this. And which, uh, which by the way, leads me to my first lesson, which is anytime there's, like, random yak shaving, I think one underappreciated, um, uh, benefit of agents is that they save you the yak shaves.

[4:24](https://www.youtube.com/watch?v=zepu8Kk6FBQ&t=264s) Like, all the dependency tree crawling of like, "Oh, no, I have to do that first. Oh, no, I have to do that first." And particularly when it comes to installing dependencies or fixing Python dependencies, fantastic for that.

[4:34](https://www.youtube.com/watch?v=zepu8Kk6FBQ&t=274s) And I think, um, a, a mo- model of productivity that doesn't sufficiently appreciate parallelism and not just autonomy, I think, uh, and, and sort of depth of the yak shaving is not fully capturing the, the benefit of agents.

[4:46](https://www.youtube.com/watch?v=zepu8Kk6FBQ&t=286s) Um, so anyway, back to the agent story. Uh, hooked up Devin to Figma, and we, uh, in very short order, we have a perfectly functioning website, uh, that is pixel perfect to the Figma.

[4:57](https://www.youtube.com/watch?v=zepu8Kk6FBQ&t=297s) And to me, uh, that was a surprise because I'd never done it before. You know, you always mistrust marketing until you see it for yourself. And more importantly, our, our designer was very happy about it.

[5:07](https://www.youtube.com/watch?v=zepu8Kk6FBQ&t=307s) Um, and that's the, that's basically the, the website that you see live today when you go to ai.engineer. Um, the other interesting thing that happened was then we started using it more, right?

[5:17](https://www.youtube.com/watch?v=zepu8Kk6FBQ&t=317s) After, after one initial success, you start using it more. Uh, something that you can't see because it's very small text, but I'm gonna highlight for you, is that that is two hundred and seven replies just exploding in usage.

[5:27](https://www.youtube.com/watch?v=zepu8Kk6FBQ&t=327s) Like, what the hell? Um, and when you dig into it, um, it's, it's very interesting, right? So first of all, uh, I start kicking off some, some work, and then I go to bed, and then, uh, my designer who's in Indonesia wakes up and starts messing with Devin.

[5:41](https://www.youtube.com/watch?v=zepu8Kk6FBQ&t=341s) Starts prompting Devin with red lines, uh, on annotations, which is something that, uh, Steve Ruiz, one of our speakers from yesterday, does with tldraw. And I never taught him to do this, and there's no instruction manual.

[5:52](https://www.youtube.com/watch?v=zepu8Kk6FBQ&t=352s) It was just mostly like, how would you communicate with another human being? And so I work mostly with a non-technical team, and I think that's very important that they need to be comfortable with agents, and I think they're finally at the point that they are.

[6:04](https://www.youtube.com/watch?v=zepu8Kk6FBQ&t=364s) Um, we start working on things that we would never normally have worked on. Uh, nobody has reported this, so I assume none of you have discovered it, but there's a Easter egg on the website.

[6:12](https://www.youtube.com/watch?v=zepu8Kk6FBQ&t=372s) Why? 'Cause I put it there. Why? 'Cause it was fun. 'Cause I could, right? So if you're on an ultra wide, you sc- you scan your mouse over the, the highlights, you'll, you'll see an Easter egg.

[6:21](https://www.youtube.com/watch?v=zepu8Kk6FBQ&t=381s) Um, uh, I saw a tweet that was viral a- about a design aesthetic that I liked. I threw it into Devin. Out pops, um ...

[6:32](https://www.youtube.com/watch?v=zepu8Kk6FBQ&t=392s) And then, and then, and then, you know, 127 replies later, uh, I, I, literally I popped it in. I was like, "Oh, let's just see what the, the clanker will, will do for me.

[6:39](https://www.youtube.com/watch?v=zepu8Kk6FBQ&t=399s) Uh, I don't want to waste my designer's time. I just wanna see what a clanker do, does for me." Designer jumps in,

[6:45](https://www.youtube.com/watch?v=zepu8Kk6FBQ&t=405s) um, and, and does, and does, and start, actually starts working on this thing which I thought was throwaway and fun. And the most interesting thing, it's so small I can't even read it.

[6:53](https://www.youtube.com/watch?v=zepu8Kk6FBQ&t=413s) I'm so sorry for this. Uh, so basically the reason he starts working on it, even though it's a throwaway project, is because it's fun, and I think that's something, uh, that was like a big aha moment for me.

[7:02](https://www.youtube.com/watch?v=zepu8Kk6FBQ&t=422s) Like, I am getting more work out of my employees because they enjoy doing it, because the feedback cycle for them from, like, waiting, blocking on me or a contract d- uh, developer that we have is gone.

[7:13](https://www.youtube.com/watch?v=zepu8Kk6FBQ&t=433s) Like, they, they just literally, they have the idea, they go do it, right? Um, and, uh, they're doing more things. They're doing animations, they're doing polish. Uh, things that we've just- I'm getting work that I've never gotten out of, uh, my employees before, and I think that's something that's appre- that's something you, you should appreciate, too.

[7:29](https://www.youtube.com/watch?v=zepu8Kk6FBQ&t=449s) Um, I'm-- If you haven't noticed, I'm no longer talking about agents for coding or, like, how many lines of code I'm producing. I'm getting more productivity out of my humans, and I, I, I think this is something that is a major theme for this year that I'm really trying to investigate, which is agents for everything else.

[7:45](https://www.youtube.com/watch?v=zepu8Kk6FBQ&t=465s) Um, then obviously, okay, I had that success with Figma to, uh, to, to website. I have the success with tweet to website. What else? Right? Like, you start to think about other use cases.

[7:57](https://www.youtube.com/watch?v=zepu8Kk6FBQ&t=477s) Um, this whole conference is a giant data management problem. Like, I have to sync with 130 speakers and a couple dozen sponsors and all the attendees that come in with all various, uh, needs.

[8:11](https://www.youtube.com/watch?v=zepu8Kk6FBQ&t=491s) Um, and really, uh, it's just a CMS, right? Like, we, we, we've, we've messed with the Sanity. I'm not a, the, the biggest fan of Sanity in the world, um, because I want to keep some sanity to myself.

[8:22](https://www.youtube.com/watch?v=zepu8Kk6FBQ&t=502s) Um, but, but basically, like, I, I can throw in, uh, spreadsheets and, and Devin can manage that for me. And once I really-- I think the unlock ho- uh, happened when I threw away the CMS and just kinda committed that to code, but used that code as my s- sort of source of truth, and let Devin, whatever

[8:36](https://www.youtube.com/watch?v=zepu8Kk6FBQ&t=516s) coding agent you use, uh, start to manage it. And so this entire schedule, uh, is managed by Devin. What does that mean? It means that whenever, um, someone comes in with a speaker change, for example, Marta, one of the speakers from today, uh, sends in an email, I just say, "Devin, handle it for me," right?

[8:54](https://www.youtube.com/watch?v=zepu8Kk6FBQ&t=534s) No other furth- further communication is needed. I can just forward the email. I can, uh, paste a screenshot, whatever. Um, and that kind of volume lets us as a small team of nine people manage a thousand-person conference, right?

[9:06](https://www.youtube.com/watch?v=zepu8Kk6FBQ&t=546s) We're gonna manage 6,000 people in San Francisco this fall, uh, th- this summer. Um, and w- I'm pretty sure we can stay the same size. Like, it, it is incredible the amount of productivity that you can get once you're sufficiently onboarded and you have the workflows ironed out.

[9:20](https://www.youtube.com/watch?v=zepu8Kk6FBQ&t=560s) Um, we have agents for ETL. We, we deal with an external vendor system that has data that we don't have in a, in a central source of truth, so I need to get the API key to sync over data and make sure there's a single source of truth.

[9:33](https://www.youtube.com/watch?v=zepu8Kk6FBQ&t=573s) Uh, these are very boring routine tasks. Um, well, there's, there's another, you know, another fun story that I can tell you is agents for buying. Uh, so I saw this viral tweet about how somebody put a claw in, uh, Wall Street next to the Wall Street bull, and I was like, "Oh, that's funny.

[9:47](https://www.youtube.com/watch?v=zepu8Kk6FBQ&t=587s) Like, we should put a claw in front of our, um, conference," and that's exactly... And so, so I asked Devin to research, "Where can I get a lobster in London?"

[9:58](https://www.youtube.com/watch?v=zepu8Kk6FBQ&t=598s) Devin comes back with phone numbers and email addresses and websites, and I just click through and, and think about it and ask it to do some more research, uh, and out pops this guy.

[10:08](https://www.youtube.com/watch?v=zepu8Kk6FBQ&t=608s) Uh, that's literally the, the lobster that you had was bought from Devin. Uh, and I think, um, this kind of personal automation for everything else, it just matters that you have an agent that has web access, that has some, uh, smart enough model.

[10:22](https://www.youtube.com/watch?v=zepu8Kk6FBQ&t=622s) Uh, it-- I mean, this is effectively a claw, right? Like, a, an OpenClaw, NanoClaw, whatever the, whatever clanker you call it. It doesn't really matter. It matters that you're using agents for things that you would otherwise have spent knowledge work on.

[10:34](https://www.youtube.com/watch?v=zepu8Kk6FBQ&t=634s) I might have had an executive assistant, I might have had a junior employee do these things for me, but now I can do it serverless on demand with a coding agent.

[10:44](https://www.youtube.com/watch?v=zepu8Kk6FBQ&t=644s) Um, I'm not here to only show Devin. Uh, I, I, you know, I o- I just advise with the company now. Um, but, uh, I, you know, I started exploring Town because I think, uh, what's, what's happening here is coding agents kind of breaking containments, right?

[10:57](https://www.youtube.com/watch?v=zepu8Kk6FBQ&t=657s) Um, there's all these other more f- fit for purpose knowledge management tools, uh, like the wikis that Andrej Karpathy is, is talking about that, uh, NanoClaw, uh, that OpenClaw is, is, is now adopting as well.

[11:08](https://www.youtube.com/watch?v=zepu8Kk6FBQ&t=668s) Um, you're gonna see an explosion of this this year. Uh, this is, like, probably the top trend of maybe top three to five trends, uh, of 2026 that I want to alert you to.

[11:17](https://www.youtube.com/watch?v=zepu8Kk6FBQ&t=677s) Um, so here is me managing t- uh, the World's Fair in 2026, uh, in, in, in this summer. Uh, here are all the tracks I'm planning. Here's my Apple Notes.

[11:27](https://www.youtube.com/watch?v=zepu8Kk6FBQ&t=687s) Uh, on the left is my Apple Notes of all the people. I'm, uh, it's intentionally small, and I threw it into, into Town, and on- out pops a nicely formatted Notion doc with, uh, research on all the speakers that I intend to, uh, solicit and, uh, think about curating.

[11:43](https://www.youtube.com/watch?v=zepu8Kk6FBQ&t=703s) Um, and then obviously once you get enough psychosis, you are thinking about replacing entire pieces of SaaS. Here's me arguing with my employees [laughs] about kicking out a, a SaaS tool and building it ourselves because we can.

[11:57](https://www.youtube.com/watch?v=zepu8Kk6FBQ&t=717s) Um, so I clearly have the most psychosis. I think one of the annoying things is if you're in a position of power or management to deal with employees who are, who are not as much in psychosis and try to bring them along the journey and, and not, uh, talk down or, or, or, or ignore their concerns, right?

[12:13](https://www.youtube.com/watch?v=zepu8Kk6FBQ&t=733s) Because they are very valid concerns, because they are exactly the people that will have to deal with your bullshit when you get it wrong, and we do get it wrong.

[12:20](https://www.youtube.com/watch?v=zepu8Kk6FBQ&t=740s) Um, so, uh, one, one ch- one top, one method I, I, I'm approaching these sort of AI replacing SaaS concept, which I think is, is, should be relevant for a lot of you, uh, is, well, let's identify the top three concerns and let's systematically reduce, uh, reduce them.

[12:34](https://www.youtube.com/watch?v=zepu8Kk6FBQ&t=754s) And that's the process that we're going through right now. So, um, I just wanted to give you a little bit of that t- taste of like, here's how AI is changing our business, uh, as managing the conference.

[12:44](https://www.youtube.com/watch?v=zepu8Kk6FBQ&t=764s) Um, it's come really, it's come really a long way. I- it's a, it's a consistent theme I'm seeing even among our, our speakers. Uh, this is Malte opening keynote, uh, talking about how the, the 60% of the, the, the sort of user base of Vercel now is bots, is, is, is agents.

[13:00](https://www.youtube.com/watch?v=zepu8Kk6FBQ&t=780s) It's not humans. So actually, your dashboards don't matter. Your APIs matter, your CLIs matter, your MCPs matter. Um, here is the MCP, uh, apps guys, Ido and Liad, uh, who spoke today, um, on, speaking on ETN about how basically your custom UI is kind of going away.

[13:17](https://www.youtube.com/watch?v=zepu8Kk6FBQ&t=797s) Like, you should ship UI to, uh, somebody else's app. And I think, like, this patterns of, like, how your primary user is changing is really shifting towards what people are calling agent experience, and I think that's something that, again, I'm really inspired by and focused on because it is helping me, right?

[13:32](https://www.youtube.com/watch?v=zepu8Kk6FBQ&t=812s) I no longer care about the Figma dashboard. I throw it into co- Claude Cowork and I hope that it works for me. Um, so that's my message. Agents for everything else are coming.

[13:42](https://www.youtube.com/watch?v=zepu8Kk6FBQ&t=822s) Wake up, use it, bring it home to work. If people are insufficiently bought in, prescribe them one of these. Thank you. [clapping] [upbeat music]
