Director of Product @CopilotKit. Author of Agentic GraphRAG (O’Reilly). Professional gate-opener. Self-improvement at DeveloperMicroskills.com ✉️.
PNW
- Tweets 27,114
- Following 3,025
- Followers 10,080
- Likes 41,459
Pinned Tweet
Thinking a lot about what @GergelyOrosz calls "product-minded engineering" & the recent work of folks like @kentcdodds.
Here are my thoughts from the product leadership side of the table & what I've learned from industry greats like @ttorres & @shreyas:
1
2
17
3,660
Sam Julien retweeted
Microsoft just released .NET SDK for AG-UI
Now any software that runs on .NET can plug C# agents into web, mobile, chat apps and more on the same open standard.
→ contributed to AG-UI
→ built next to Python, TypeScript
→ Agent Framework relies on it
Walkthrough + code from @danroth27 and the @dotnet team: devblogs.microsoft.com/dotne…
4
3
11
563
How do you write product requirements for an AI agent?
“Add an assistant” leaves a lot unanswered. When does it ask permission? How can users change their minds?
AG-UI gives PMs a concrete vocabulary for designing how people and agents work together.
6
2
10
998
Do we publish and tweet on Saturdays these days? Is that a thing or should I just wait until Monday?
5
356
Sam Julien retweeted
Jev 🤝 CopilotKit
We tested UI rendering side by side to demonstrate how Jev from @typesafeai can instantly stream components with GPT-6 Luna.
Same prompt, renderer and sample data.
5
21
172
21,107
The weekly agentic product loop:
1. Research & discovery
2. Strategy refinement
3. Pricing & packaging
4. Prototyping
5. Scoping & slicing
6. Implementation & release
7. Validation & testing
♻️♻️♻️
2
4
346
The hardest part of being in product right now is being able to keep up with the pace of production. I need to be scoping and spec’ing high impact work on the daily.
1
4
1,814
We’ve figured out that, almost regardless of the engineering scope, a new real feature takes ~3 days to get to prod at scale:
- 1 day of grilling and looping
- 1 day of validating and fixing
- 1 day of releasing and testing
(Not applicable to frontend only)
You finding this?
7
9
1,119
Trading scope with engineering leads in days instead of weeks for complex features is still mind-blowing to me.
Me: “Okay so we’ll have [incredibly complex new feature] by Wednesday?”
Them: “Whoa whoa whoa Wednesday?!
I can do Thursday or Friday.”
6
376
O’Reilly sent me author copies of the book and someone smashed open the box in transit, stole the books, and swapped in two random paperbacks 😭
Enjoy the GraphRAG I guess?!?
4
13
1,046
Sam Julien retweeted
It’s live on Product Hunt 🔥
Agentic developer marketing system for technical founders who think in systems.
producthunt.com/posts/quiver…
14
3
31
776
Hoo boy someone just asked to schedule something in December 😮💨 are we really at that point already??
1
314
It's wild to me that if I have to do something like wait at a car dealership, it's actually more efficient for me to leave my laptop at home so that I can remote in with Codex than it is to bring it with me because of the poor Wi-Fi
1
4
612
Sam Julien retweeted
New in OpenMuse: Automatic Learning
The more your OpenMuse is used by users, the better it gets. Product trajectories from user-interactions are turned into skills for your agents.
Try it: github.com/CopilotKit/openmu…
𝙾𝚙𝚎𝚗𝙼𝚞𝚜𝚎 now has Automatic Learning 🌟
→ Build OpenMuse with any agent/frontend
→ Intelligence surfaces insights & skills
→ Approved skills get ingested into agent
So the more your OpenMuse is used, the better it gets.
Powered by @CopilotKit & AG-UI.
Try it: github.com/CopilotKit/openmu…
4
16
104
9,904
Can confirm @HamelHusain and @sh_reya are the eval GOATs. Took their course last year and it fundamentally changed how I build AI products & platforms. I also cite them in my evals chapter in my O’Reilly book.
10/10 would look at my data again
Evals have been coming up more and more in my conversations with podcast guests and PM friends.
Nearly half of the 25 awesome PM job openings I shared last week ask for experience writing evals. And leading companies keep sharing what investing in evals bought them:
— @tryramp took its automatic receipt collection from 35% to 83% accuracy.
— @Shopify shipped an AI workflow builder that's 2.2x faster and 68% cheaper than the frontier-model setup it replaced.
— @harvey__ai rebuilt its AI contract reviewer, nearly doubling its internal quality score.
— @cursor_ai tuned its Auto Balance routing, with much higher user satisfaction at 41% lower cost.
So I asked the 🐐s of evals, @HamelHusain and @sh_reya, to write an advanced sequel to their very popular "Building eval systems that improve your AI product." Drawing on their work with 50+ AI companies, they share the key step most teams skip, what you should (and shouldn't) automate, and a free plugin that lets a coding agent do most of the heavy lifting.
Read it here: lennysnewsletter.com/p/advan…
1
4
17
17,033
So we built OpenMuse! 🎉
But here's something even crazier: OpenMuse can learn from your user interactions thanks to CopilotKit and AG-UI.
- Build OpenMuse w/any agent/frontend
- Intelligence learns and proposes skills
- Approved skills automatically ingest into agent
Try it!
🎉 Introducing 𝙾𝚙𝚎𝚗𝙼𝚞𝚜𝚎
An open source, self-hostable personal assistant that works with any agent harness.
Includes:
- Computer use: browser, terminal & files
- Connectors for your personal apps
- Ideas, goals & progress tracking
- Built for Mobile and Web
Repo → github.com/CopilotKit/OpenMu…
Powered by @CopilotKit and AG-UI.
Clone this template and customize it however you want.
5
18
2,816
Sam Julien retweeted
🎉 Introducing 𝙾𝚙𝚎𝚗𝙼𝚞𝚜𝚎
An open source, self-hostable personal assistant that works with any agent harness.
Includes:
- Computer use: browser, terminal & files
- Connectors for your personal apps
- Ideas, goals & progress tracking
- Built for Mobile and Web
Repo → github.com/CopilotKit/OpenMu…
Powered by @CopilotKit and AG-UI.
Clone this template and customize it however you want.
Introducing Muse, your personal AI agent from Meta that gets things done across every part of life.
Download the Muse app and get started: Muse.ai
298
512
5,856
1,093,106