I spend a lot of time on AI, agents, developer tools and how software is changing. I share what I’m building, learning and finding useful along the way.
Joined July 2010
- Tweets68
- Following428
- Followers98
- Likes15
Really liking Capy’s Captain 🐹
One chat to plan + steer a crew of agents working in parallel — without jumping across threads. This is how multi-agent coding should feel.
Try it: capy.ai/?ref=01M3WXHP07GV3TX…
Thanks @justinsunyt for the credits 🙏
And being able to use your existing Codex subscription is 👌
For heavy agent usage, this matters a lot. You can use the subscriptions you already pay for rather than treating every agent run as another separate token bill.
Jev is moving too fast to track, so I built a live radar of what people are shipping entirely from my phone in a few hours with @AmpCode.
@typesafeai’s Jev scores each post for novelty, leverage & builder value.
👉 jevradar.com
What should I add next?
Jev by @typesafeai isn’t a smaller LLM.
Chat models generate. Jev judges.
state + question → probability → code
That’s why it fits routing, ranking, gating, agents, not chat.
A good read to understand more about Jev: jevradar.com/jev
Jev by @typesafeai isn’t a smaller LLM.
Chat models generate. Jev judges.
state + question → probability → code
That’s why it fits routing, ranking, gating, agents, not chat.
A good read to understand more about Jev: jevradar.com/jev
What if every if statement could understand meaning?
That’s what @typesafeai’s Jev makes interesting.
10 things I’d build with it 🧵
10. Build a probabilistic state machine.
Traditional software has rigid states.
Real-world workflows are fuzzy.
ready_to_ship? → .91
needs_review? → .08
customer_risk? → .11
requirement_complete? → .96
Meaning becomes part of program state.