Creator of the AI Unified Process. Java Champion, Vaadin Champion, Oracle ACE Pro, Author, Speaker, Programming Architect, and Lecturer.
Erlach, Schweiz
- Tweets 28,782
- Following 1,681
- Followers 10,016
- Likes 22,684
Pinned Tweet
Surprise! My book arrived two weeks before the official release date. 🎉
"Spec-Driven Development: From Specs to Code with AI Agents" (Apress Pocket Guides) is now available, both as softcover and eBook.
It is a compact guide on how to make specs the contract between you and your AI agents, based on the AI Unified Process (AIUP). Requirements, use cases, entity models, and a complete case study, all in about 150 pages.
I hope it helps you bring more structure into AI-assisted development.
👉 link.springer.com/book/10.10…
#SpecDrivenDevelopment #AIUnifiedProcess #GenerativeAI #KeepITSimple
8
32
232
14,200
From Developer to Builder: A Spec Driven Development Deep Dive
The recording from Devoxx 2026 is available on YouTube.
piped.video/watch?v=z8lal-Yt…
2
317
Says @simas_ch: “Write the use case before you prompt: write the actors, the main scenario, the alternative flows & the business rules in a simple Markdown file. Give that file to your AI agent and let it implement the feature and the tests from it."
🔗 ow.ly/4rR550ZR08w
1
4
254
I highly recommend this conference 👇
I had the privilege to be a speaker at the London edition this spring.
London sold out. NYC is next.
AI DevCon New York, 2–4 Nov, Brooklyn: 700+ engineers on agents after the demo.
On the agenda:
→ @GoogleDeepMind: stop babysitting your agents
→ @SpaceXAIMemphis: PRs written end to end by agents
→ @datadoghq: is your agent rollout working?
→ @temporalio : when agents run their own tools
Workshop seats are limited: aidevcon.io
3
315
What if your specs became the foundation for code, tests, and validation? Learn from @simas_ch how a spec-first approach reduces guesswork, improves alignment, and helps teams build software more efficiently. #SoftwareDevelopment #DevOps #Engineering
🔗 ow.ly/evz250ZW0HX
1
2
227
Simon Martinelli retweeted
Long ago it was a developer rite of passage to create a web framework. Last year creating an agent framework was the rite of passage. This year it is agent harness. Get out there and build yourself a harness! You’ll learn a lot and might even create something exceptional.
12
1
31
2,293
Vaadin is AI native. And it's an amazing fit for Spec-Driven Development.
Vaadin 25.3 is built for AI coding agents that work on Java UIs:
✅ A dev loop that works like an IDE for your agent
✅ A self-hosted MCP server with the full Java API (even air-gapped)
✅ Browserless tests, so the agent gets fast feedback without a browser
✅ An Observability Kit insights endpoint your agent can read
Why does this matter for Spec-Driven Development?
With Vaadin, the UI is plain Java. No JavaScript, no REST layer, no second tech stack. The agent works in one language and one codebase. That makes it much easier to go from the spec to running code, and to check with tests that the code still matches the spec.
Specs give the direction. Vaadin keeps the path short.
🎥 Watch the 5-minute overview: piped.video/watch?v=WnkI6xA9…
📣 Want to see it live? Next week I speak at Vaadin Create 2026 in Barcelona (27-28 October): "Spec-Driven Development - Why Vaadin Is the Perfect Fit"
AI can generate code. The hard part is to keep it aligned with the system you really want to build. I will show how specifications and architectural guardrails guide a full-stack Java application while it grows.
See you in Barcelona! 🇪🇸
vaadin.com/vaadin-create-202…
#Vaadin #Java #SpecDrivenDevelopment #AIUP #AI #VaadinCreate
3
7
20
1,207
Leanpub Book LAUNCH 🚀 Writing Use Cases for AI by Simon Martinelli #book... piped.video/O9mdf34ibD0?is=zNkX…
1
11
678
Simon Martinelli retweeted
If you silently correct the mistakes you junior coworkers makes, without telling him what was wrong and why it was wrong, you deprive them from learning and they're going to make the same and similar mistakes again.
Same with your AI coding agent.
2
2
3
250
Postgres Just Went Realtime | Neon Changelog piped.video/shorts/6HQrNMQ8P…
4
698
Says @simas_ch: “Write the use case before you prompt: write the actors, the main scenario, the alternative flows & the business rules in a simple Markdown file. Give that file to your AI agent and let it implement the feature and the tests from it."
🔗 ow.ly/4rR550ZR08w
2
4
313
I never liked the idea of PRs in business applications. Overall reviews are way more accurate to share knowledge. Which is also important in times of AI
Why are you still using pull requests?
Is it for code review?
If so, do you really think your or your colleagues catch bugs and mistakes that an Opus 5.5 or Fable or Astra won't find? You can literally spawn all three of them and have them check four times each.
(Also, don't tell me you don't have a "stamp plz" Slack channel or group chat somewhere.)
Is it for knowledge sharing?
Surely there's better ways for that now than the PR, no? We have models that can generate explainer videos, presentations, podcasts, games, quizzes, walkthroughs, ... for any piece of code it can reach.
Is CI the reason?
Well ackshually, CI means continuous integration and its original meaning is the *opposite* of having long-lived branches and PRs.
But, leaving that aside, your agents are hopefully already running the tests before they push something, and hopefully they can run as long as they want and don't block you so they can run the whole test suite and generate "irrefutable proof" and end-to-end demo videos and so on... And so they run the tests and confirm and then you push to a branch and everything runs again and then you merge and then it runs again? That doesn't make sense, does it?
Maybe you work in open source and you accept contributions, in which case I can't speak to the value of pull requests. It's been 2 years since my last OSS contribution. But from what I heard, the time of the PR is coming to an end in OSS too?
I honestly don't get why people still stick to PRs.
With agents you can now move 10x faster, you can finally do real continuous integration, and get more confidence in the changes the agent made.
Don't slow yourself down by having all the changes pile up in a PR.
9
1
13
1,519
Simon Martinelli retweeted
Why are you still using pull requests?
Is it for code review?
If so, do you really think your or your colleagues catch bugs and mistakes that an Opus 5.5 or Fable or Astra won't find? You can literally spawn all three of them and have them check four times each.
(Also, don't tell me you don't have a "stamp plz" Slack channel or group chat somewhere.)
Is it for knowledge sharing?
Surely there's better ways for that now than the PR, no? We have models that can generate explainer videos, presentations, podcasts, games, quizzes, walkthroughs, ... for any piece of code it can reach.
Is CI the reason?
Well ackshually, CI means continuous integration and its original meaning is the *opposite* of having long-lived branches and PRs.
But, leaving that aside, your agents are hopefully already running the tests before they push something, and hopefully they can run as long as they want and don't block you so they can run the whole test suite and generate "irrefutable proof" and end-to-end demo videos and so on... And so they run the tests and confirm and then you push to a branch and everything runs again and then you merge and then it runs again? That doesn't make sense, does it?
Maybe you work in open source and you accept contributions, in which case I can't speak to the value of pull requests. It's been 2 years since my last OSS contribution. But from what I heard, the time of the PR is coming to an end in OSS too?
I honestly don't get why people still stick to PRs.
With agents you can now move 10x faster, you can finally do real continuous integration, and get more confidence in the changes the agent made.
Don't slow yourself down by having all the changes pile up in a PR.
416
85
1,696
390,694
Thanks to the excellent Design Systems from @vaadin which are available on Figma design UIs using the Figma MCP is a breeze.
I use Claude Code to create the design for a use case, then discuss that with my customer, adjust if needed, and advise Claude Code to implement the use case according to the design.
figma.com/@vaadin
5
3
21
902
Next UX fail: @Microsoft Teams!
I changed the password of a M365 account. The mobile Teams app keeps showing old messages and no notification to re-logon. 🤷♂️
3
6
826
A lot of products have bad UX.
Latest example I was struggling with is Bitbucket. Why is it so hard to disable an option when it's obvious that this option cannot be used?
The most annoying thing in Bitbucket and also in GitHub is: you merge a PR then go back to the PR list and the PR is still listed as open.
Please improve the UX @Bitbucket and @github
1
3
660
What if your specs became the foundation for code, tests, and validation? Learn from @simas_ch how a spec-first approach reduces guesswork, improves alignment, and helps teams build software more efficiently. #SoftwareDevelopment #DevOps #Engineering
🔗 ow.ly/evz250ZW0HX
2
5
257
Good article from Vaadin about the “AI control gap”: the space between your prompt and the code you accept.
I fully agree with “control by construction”: one language, one compiler, current context through MCP and skills, fast tests.
But one thing is missing. Context tells the AI how to build. It does not tell the AI what to build.
Control starts with the specification. Requirements, use cases, and business rules are the source of truth. Tests trace back to them. Then the compiler, the MCP server, and the tests have something clear to check against.
That is what Spec-Driven Development and the AI Unified Process are about.
Vaadin + Spring Boot + jOOQ + a good spec: that’s my setup for keeping control.
#Java #Vaadin #AI #SpecDrivenDevelopment #AIUP
In Veracode's 2026 tests, Java had the lowest security pass rate for AI-generated code of any language: 30%. It was also the only language clearly improving.
Setup matters more than the assistant. Six habits for Java teams:
vaadin.com/blog/ai-coding-fo…
ALT Feature image titled 'AI coding for Java teams: speed without losing control'. A prompt leads into a dashed box labeled 'AI control gap', followed by a card of Vaadin Java code.
9
1
17
726
Interesting but be careful! You probably don't have the required for version control that Anthropic has.
New on our Engineering blog: Creating in-house source control.
In May, we merged around 2,500 code changes per day. In September, we merged more than 30,000 per day. As a result, we became bottlenecked by developer infrastructure built for human (rather than agent) scale.
Here, we share how we built a new version control system—and fully migrated our monorepo from git—in five months: anthropic.com/engineering/re…
2
1
6
705