Laura Summers

You can now trivially build coding agents and general knowledge agents with @pydantic AI! Like pi, but in Python. Codepuppy showed this was possible, and has seem explosive adoption within enterprise. Now we've made those agents easy to compose with just Pydantic AI. @openclaw

Starlette (and uvicorn) are cornerstones of the Python networking stack.

Starlette 1.0 is here!🎉 After nearly eight years, Starlette has reached its first stable release. Downloaded almost 10 million times a day, it serves as the foundation for FastAPI and the Python MCP SDK. Blog post: marcelotryle.com/blog/2026/03/2… Release notes: #100-march-22-2026" target="_blank" rel="nofollow noopener">starlette.io/release-notes/…

Starlette 1.0 is out! I used this as an opportunity to experiment with Claude Skills, since Claude isn't yet familiar with the (minor) breaking changes in the 1.0 release compared to 0.x simonwillison.net/2026/Mar/22/st…

great read on the state of oss and ai pydantic.dev/articles/pyai-…

Thoughts on OpenAI acquiring Astral and uv/ruff/ty simonwillison.net/2026/Mar/19/op…

AI is making CEOs delusional

Programming with LLMs is genuinely useful and genuinely destabilising. Our lead designer @summerscope wrote about what it actually feels like to be a developer right now. If you're feeling simultaneously more productive and less satisfied, you're not broken. The feedback loop is. Full story on our blogpost pydantic.dev/articles/the-h…

We shipped Claude Code as a research preview a year ago today. Developers have used it to build weekend projects, ship production apps, write code at the world's largest companies, and help plan a Mars rover drive. We built it, and you showed us what it was for.

Find me a conference with a better line-up, I dare you. There are still some tickets available, buy now. pyai.events/speakers Or, even better, if you're a serious open source developer in your free time, email me and we'll give you a free ticket. samuel at pydantic.dev.

Logfire is getting a UI refactor. That means a lot of docs pages with screenshots that no longer match. Instead of updating them manually, we used @anthropic's Claude Code + @vercel's agent-browser to automate it. It opens the UI, clicks through workflows, captures screenshots, reads the existing Markdown, and updates everything in place. @danielcruzdev wrote up the details on our latest blogpost pydantic.dev/articles/autom…

Introducing: built-in git worktree support for Claude Code Now, agents can run in parallel without interfering with one other. Each agent gets its own worktree and can work independently. The Claude Code Desktop app has had built-in support for worktrees for a while, and now we're bringing it to CLI too. Learn more about worktrees: git-scm.com/docs/git-workt…

Today's Operator Links: operatorlinks.com/2026-02-20 1️⃣ Creation Is Infinite Now. That's the Problem. - Elliot Bonneville 2️⃣ 10 Years. Hundreds of Projects. Still No Discord. - James Liu 3️⃣ You Can Start Five Things. You Can Still Only Finish One. - @summerscope 4️⃣ Stop Building AI That Decides. Build AI That Lifts. - Ben Gregory 5️⃣ AI Agents Can Handle Five Hours. We Give Them 45 Minutes. - Anthropic

@zeeg Come to our conference. 😉 pyai.events

@samuelcolvin Did you reset your effort to `/medium`? high effort is really slow

Is it me, or is Claude Code extremely slow these days? Not necessarily a bad thing, maybe it'll think for a very long time and just come back with a solution - this is a relatively hard problem. But it definitely seems significantly slower than pre 4.6 or immediately after the 4.6 release.

We used Logfire to trace what really happens during a @FastAPI request and measured where time is actually spent. Read about it on our latest blogpost. pydantic.dev/articles/web-f…

How can I overcome the "smile-or-die American corporatism tone in LLM responses? This was a great question from @summerscope (@pydantic) last year in our Building with AI course! Tone in AI apps is an fundamental design choice, not always easy to get right, and there are several strategies you can employ (we also discussed how to evaluate them!) Our final cohort starts March 10. 25% off with code FINAL25: maven.com/hugo-stefan/bu…

@BlakeOperator @hugobowne @pydantic Interesting- I’ve been experimenting with specialised reviewer agents but hadn’t thought to address text tone like this. Definitely worth a try.

@hugobowne @summerscope @pydantic The "toxic positivity" in default LLM weights is a real productivity killer. I’ve found that the only way to kill it isn't better prompting—it’s Phase Separation. One agent to reason, and a cold "Extractor" agent to strip the "I hope this helps!" fluff. 🛠️

We are thrilled to announce @mitsuhiko as a speaker at PyAI! If you've built anything in Python, chances are Armin's work is in your stack. He's the creator of Flask, Jinja, Werkzeug, Click, and more. Tools that millions of developers rely on every day. Join us in San Francisco. Co-hosted with Prefect. Grab your ticket today! pyai.events

Pydantic just hit 10 billion downloads. Born out of @samuelcolvin's frustration in 2017. Nine years and some breaking changes later: we became a company, shipped Logfire, released Pydantic AI, and recently released Monty. Full journey on the our blog. pydantic.dev/articles/pydan…