About
A radar for people who build businesses, not just software.
AI is merging engineering and product management. One person with good agents can now take a customer problem all the way to a working business. The bottleneck has moved from the craft of building to finding problems, shipping fast and turning products into businesses.
Who it's for
- Solopreneurs and indie founders who build and run products largely alone, with AI.
- Product managers whose role now includes building, because agents do much of the implementation.
- Software engineers moving towards product, user craft and design as agents take over more of the coding.
The roles are converging. All three care about leverage, speed and outcomes more than a tool's technical depth.
How to read the radar
Each entry is a blip. Its quadrant says what kind of thing it is. Its ring says what to do about it. The closer to the centre, the stronger the recommendation.
- ADOPT
- Proven in real use. A default choice.
- TRIAL
- Worth using on real work where the risk is manageable.
- ASSESS
- Worth understanding and watching.
- HOLD
- Avoid for new work, or superseded.
Colour is for change
Blips that did not move this edition are drawn in plain ink. Colour marks only what changed, so the eye goes to movement first. A moved blip leaves a trail back to the ring it came from, like target history on a radar scope.
- New Added this edition.
- Moved in Moved towards the centre: a stronger recommendation.
- Moved out Moved towards Hold: a weaker recommendation.
- Retired Left the radar this edition; listed under What moved.
- No change No change this edition.
Quadrants
- Models & Agents
- Foundation models and the agents built directly on them: coding, voice, image, reasoning.
- Tools & Harnesses
- Apps and CLIs you work through: agent harnesses, business tools, CRMs, knowledge bases.
- Infrastructure
- Hosting, deployment, data, agent-friendly platform interfaces and local model runtimes.
- Practices
- Techniques and workflows: how to build, validate and run a business with agents.
Signal over noise
The space produces more launches than anyone can track, and a radar that lists everything is useless. An entry earns its place only if it changes what one AI-enabled person can build, ship or run. Novelty alone isn't enough.
The radar is curated by one practitioner, based on what is actually used to build and run real businesses, plus what is moving in the wider market. It is first-hand and opinionated. It is reviewed roughly once a week: entries are added, moved or retired, and every change records why.
From one radar to many
This starts as one curated radar. The longer-term idea is that anyone can keep a radar of the stack they actually use, keep it current and compare it with others. Disagreement is useful signal. Build yours or look at an example community radar.
It is deliberately not a launch site. It is about which tools people keep using, and why.
Open data
The whole radar is available as radar.json, with a plain-text summary for language models at llms.txt and every change in the Atom feed. Use it, quote it, argue with it.
Colophon
The project is also the experiment. This site is written mostly by agents, so its stack is chosen for AI-authored code: a small Rust generator whose compiler and validator catch agent mistakes, data in plain TOML files, static HTML with a little JavaScript, and hosting on Cloudflare Pages. It is built with Claude Code. Type is Archivo and Martian Mono.