AI Engineer

And that's a wrap! AI Engineer Europe 2026 has concluded. Our video crew did incredible work to capture the energy, enthusiasm, and positivity of this event -- but it still doesn't come close to being there. If you're engineering the future of AI -- we hope to see you at a future event! Thank you London -- thank you Europe!

Patrick also makes the bigger point: context is not just input, it is a flywheel. Better context -> better agent output -> better observations -> better regenerated context. Teams that learn to engineer that loop will ship faster, review less garbage, and build a real moat around their AI workflow.

Some of the most useful ideas from this talk: - Treat context as an engineering artifact, not ad hoc prompt text - Build a lifecycle around it: Generate, Evaluate, Distribute, Observe - Move from one off prompting to reusable context components - Test whether generated code actually follows the rules and starts from the right constraints - Run context checks through CI/CD, not just through human vibes - Observe context in production, then regenerate and improve it when it fails

Context may be the most under-engineered layer in AI coding today. In this keynote, @patrickdebois, argues that if agents are driven by prompts, rules, and memory, then context deserves the same rigor we already give code. youtube.com/watch?v=bSG9wU…

the clip x.com/tokengobbler/s…

🆕 Software Engineering Is Becoming Plan and Review youtube.com/watch?v=W76woO… AI eats the middle. If software engineers are spending more of their time planning work and reviewing AI output, then the biggest lever for shipping faster is improving planning and review. In this talk, @tokengobbler looks at how teams are actually adapting: where time is going, how code review is changing, what planning looks like now, and whether agile and scrum still fit the shape of AI-native software teams.

"Most people are using MCP wrong," said MCP's co-creator, David Soria Parra - @dsp_ I agree... because I made the same mistake. Last year, while building a financial assistant for a San Francisco startup, we needed Gmail, Calendar, and CRMs. MCP was trending, so we wrapped plain API calls inside a custom MCP registry. Before tools like @fastmcp by @PrefectIO, building MCP servers was painful. But even after all that… It added zero value. MCP was just a fancy wrapper around API calls. So I wrote it off completely. Then David took the stage at @aiDotEngineer and reframed the whole thing... MCP is not dead because of Skills or CLIs. We were just using it wrong. The best agents work across 4 layers: • User interfaces → Chat UI, IDEs, internal apps • Agent harness → runs the LLM + decides tool usage • Connectivity → Skills, MCP, CLI • Backends → APIs, databases, internal systems And they run on 3 legs: 1/ Skills (Domain knowledge) Simple files that teach the model how to think. They're: • Reusable •Lightweight • Easy to update 2/ CLI / Computer use (Cheap local execution) Perfect for coding agents. If the model knows git, grep, pytest, and has a sandbox, CLI is fastest. 3/ MCP (Rich semantics) • Typed tools • Resources • Long-running tasks (e.g., ETL jobs) • Auth • Governance • Platform independence. The boring enterprise stuff. The mistake wasn’t using MCP... It was using MCP as a REST wrapper. MCP servers should be products. Start from user journeys like: • Reconcile last month’s failed Stripe payouts • Approve a customer refund above $5K • Investigate why this week’s sales dashboard dropped 18% • Re-run a failed ETL job for yesterday’s finance reports Not low-level API calls. Design around domains like finance-mcp, crm-mcp, devops-mcp. Use resources + tasks. This is how you build production agents. And if you’re building MCP servers in Python, use @fastmcp by @PrefectIO In David’s words, “It’s just way f*** better than our Python SDK that we shipped.” Yes, the person who wrote the Python SDK said this. P.S. David’s full ~18-minute talk is worth watching: youtube.com/watch?v=v3Fr2J…

after @aiDotEngineer, I'm thinking on some problems from AI models to overcome, got this list from the top of my head, anything missing? • continuous learning or better memory? • long-term planning • prompt quality resilience: different prompts shouldn’t alter the outcome • pushback: they shouldn’t agree in all cases and infer the real goal • multi-step reliability • cost aware: they should know when writing a Python script is better than doing it themselves

@jancurn @swyx @AnthropicAI yup!

@swyx @aiDotEngineer @AnthropicAI @swyx time has come?

So everyone's been asking about @aidotengineer World's Fair 2025 (jun 3-5, in SF) Happy to confirm: AIE will be the first AI conference to have a dedicated MCP track, supported by @AnthropicAI. If you work on MCP we want this to be the place to meet everyone that matters. we are launching CFPs this week (put your email in the website to get notified). Lots of updates from NYC/2024 - our alpha (h/t @gwern) is that we can move the meta faster and more in tune with the AIE community than everyone else bc of the @latentspacepod community. so just to make sure i cover my bases - any other big themes i should cover with full track status?

🆕 I Gave an AI Agent the Keys to My Life (Here's What Happened) youtube.com/watch?v=sJ2jc7… A personal AI agent starts by reading files. A few months later, it is backing itself up at 2am, monitoring its own health, and drafting business emails. This talk is a real story about permission creep, trust, self-management, and what it actually feels like to run an always-on agent against your actual life.

People are really enjoying our full workshops showing end to end walkthroughs of real production workflows! This is a rare double header with @braintrust's Giran Moodley and @OussamaHaff walking though the real life AI engineering behind @thetrainline, Europe's #1 most downloaded rail app with 27m MAU and £5.3B in ticket sales! the workshop bundles several important lessons: - break down monolithic LLM calls into specialized stages (e.g., triage, policy review, and reply generation) - how to monitor latency, token usage, and costs effectively with end-to-end tracing of agentic flows - using "golden sets" (a curated set of test inputs) to identify failure modes - how to move from local development to a managed environment where prompts and scoring functions are version-controlled - how to allow non-technical team members to collaborate and update model parameters without code changes - how to identify production regressions, replay failures, and apply targeted fixes to improve system reliability continuously enjoy!

Watch here → braintrustdata.link/AI-engineer-se…

some of the parts of this event are missing attention while it is really important from @aiDotEngineer you should not really miss any, even if not going to watch it give the list/video to your assistant to scrape the ideas and check what is aligning with your vision ️

Hi Singapore 🇸🇬, meet Codex 🩵 With Codex, ANYONE can build and create. We’re turning that energy up this May. We’re a diamond sponsor 💎 at @aiDotEngineer Singapore, at a bunch of events, and hosting sessions for builders at every level. Plus, @thsottiaux will be in town. There's something for everyone: > Intro workshops if you’re just getting started. > Hack nights if you want to go deep & tokenmax > Demo nights if you want to see what people are building. May is going to be fun, and we're just getting started. +++++++++++++++++ 3 May: Emergency Codex Hackathon luma.com/zprbhtzyTogeth… with @brianchew 4 May: Codex Hack Night - Design & Frontend luma.com/bw3rpffqTogeth… with @cleondesigns 5 May: Codex Demo Night - GPT-5.5 & ImageGen 2 luma.com/tu1yajd5Togeth… with @Lorong_AI and @yongquanYQ 7 May: AI Engineer UnConference luma.com/3binn6ce No talk there, but I’ll be hanging out. Come find us. Together with the @65labslah crew 9 May: AI Engineer Singapore Hackathon Link is TBA Together with the @65labslah crew 9 May: Codex workshop at The Good Hack lnkd.in/gYrUMj3g Thanks for inviting us GoodHub SEA (Joanne Tan), Open Government Products 11 May: Codex Hack Night - Computer Use luma.com/pbk5bb32 Together with @Lorong_AI and @yongquanYQ 13 May: Codex Hack Night - TBC 👀 luma.com/4v97ndhxTogeth… Together with @Lorong_AI TBA: Intro to Codex for Everyday Work - Workshop 15 May: AI Engineer Singapore - Codex Workshop 16-17 May: AI Engineer Singapore - Keynotes + Codex Booths and more ...

What if chat is the wrong interface for managing agents? In this talk, @steveruizok shows what changes when agents move onto a shared canvas instead of staying trapped in a linear thread. Using tldraw's Fairydraw experiment, where users collaborated with three "fairies" coordinating with both humans and each other, this is a practical look at what spatial interfaces make possible for agent workflows. youtube.com/watch?v=sPUjIB…

i said on @jacobeffron's pod recently that "coding agents breaking containment" is the breakout theme of the year. i meant it - this is the year all knowledge workers, not just coders, get AGI-pilled. for the AIE EU closing note (youtube.com/watch?v=zepu8K…), I gave a short talk on how we use agents to run @aidotengineer as a Tiny Team that now serves ~1m unique developers a month for free all around the world, for everything from CMS to renting lobster inflatables. yes I use @openclaw personally and as a team we use @cognition's Devin and @townai, but this isn't about any one agent; it's about all of them, and how you are probably not trying hard enough to use them for daily knowledge work. i hope this gives you agent productivity ideas for you and your team.

we are beyond stoked to have @ryolu_ head of design at @cursor_ai speaking at @aiDotEngineer singapore! ryo is the mind behind so many of the interface decisions that make cursor feel like magic - and one of the most thoughtful voices in ai-native design right now. before cursor, he was founding designer at notion, and held design roles at stripe and asana. see you in a few weeks at ai.engineer/singapore @swyx @agrimsingh @unprofeshme @aimuggle @adlinzainal @ivanleomk

full 2026 Codex workshop here: youtube.com/watch?v=MhHEGM…

Full Workshop: @OpenAI Codex masterclass The agent is no longer just one chat window. In this workshop, @reach_vb and @kagigz get into how coding systems start to change when you can delegate work across subagents, split tasks up, and manage more context than a single thread can comfortably hold. If you're trying to understand where the next generation of coding agents is going, this is a useful look at the shape of that future from inside @OpenAI! Full Workshop: youtube.com/watch?v=MhHEGM…

🆕Building Conversational Agents youtube.com/watch?v=cVzf49… we are honored to host @_philschmid and @thorwebdev for a masterclass in building conversational agents, from tool-using coding agents to realtime voice interfaces. This full 2 hour workshop covers everything you need to know in 2026 for realtime voice/video agents, from the new Interactions API, agent skills, server-side state, and the Live API workflow for streaming audio, video, and tool calls into multimodal assistants!

@AIEMiami @lrusinou stallions

AI Engineer Miami wouldn't have been possible without our incredible volunteers. Special thank you to our Head of Volunteers, @lrusinou, for crushing it!