Why the developer-facing half of AIWatcher is open source, forever
AIWatcher Local reads your coding sessions. Here's the honest answer to why it's open source, what that commitment actually covers, and how to verify it yourself.
Nobody should install a tool that reads their coding sessions without asking why it's open source. So ask.
Here's the honest answer.
The suspicion is fair
A local agent reading your Claude Code, Codex, and Cursor sessions is not a small ask. It sees which repos you touch, how long you work, what you're building. Tools that want that kind of access and don't show you the code usually want something back later: your data sold on, a paywall that shows up after you're dependent on it, behavior that quietly changes and you'd have no way to know, because you never could see it in the first place.
So the fact that AIWatcher Local is open source shouldn't be reassuring on its own. It should be the first thing you interrogate.
The economics, plainly
Not charity. Not a loss leader with a trap door. The economics are plain, and we'd rather state them than let you guess:
Individual developers are how this company reaches the people who eventually buy the other half of the product. Local is the wedge. For a wedge to work, it has to be unconditionally trustworthy the moment someone installs it, on their own machine, with nothing asked in return. Not trustworthy after a trial period. Not trustworthy until the pricing page shows up. Trustworthy on day one, because day one is the only chance it gets.
The money is in AIWatcher Enterprise: a company choosing to bring the same control loop across its developers' laptops and the AI features inside its own product, with one team view, policy across developers and customer usage, and the intervention graph across everyone's work. That's a decision a VP Eng or Head of Platform makes, for their org, separately from whatever their developers installed for themselves.
We make money when a company adopts the loop for its product. We do not make money from you personally, and the product is built so that's structurally true, not just a promise we're asking you to believe.
What "open source, forever" actually commits to
Adjectives are cheap. Here's what we're actually on the hook for:
Local-first. The collector runs on your machine. It reads session files and databases that Claude Code, Codex, and Cursor already write locally. Nothing is uploaded, because there's nowhere for it to go: no server call in the path.
Open source. The code is public, not just the marketing. You can read exactly what the collector touches and what it doesn't.
No cost, ever. Not now, not after an upgrade prompt, not after a feature you depend on quietly moves behind a paywall.
No account. Nothing to sign up for, nothing to log into. Install and it works.
No server dependency for the core loop. If AIWatcher the company disappeared tomorrow, Local keeps working exactly as it does today. Nothing about it depends on us staying in business.
Those are five checkable claims, not five adjectives. That's deliberate.
Don't take this on faith
The right response to a company saying "trust us" is not to trust them. It's to check.
The repo is public. The part that matters most, what the collector reads, what it hashes, what never leaves your machine, is a few hundred lines you can read in the time it takes to finish your coffee. Read it. If something looks wrong, say so publicly, and we'll either fix it or explain ourselves in the open, where you can watch us do it.
That's a stronger guarantee than any policy page. A policy page is a claim. Code you can read is evidence.
The bridge is real, but it's not a condition
Some of the developers who install Local work at companies that later become Enterprise customers. That's the business model working as intended, and we're not going to pretend otherwise.
But Local's value to you doesn't depend on that ever happening. It's not a 14-day trial with a countdown hidden in the fine print. It's not going to start nagging you toward an upgrade six months in. If you install it, use it to see what your own sessions have been costing you, and never touch AIWatcher again in any other context, Local did exactly what it was built to do.
We're betting that being genuinely useful, with nothing owed in return, is what earns the second conversation. If we're wrong about that, the honest outcome is that Local stays a good tool nobody scales past, and Enterprise doesn't happen. We'd rather take that risk than build the trust on anything less than what we just told you.
See AIWatcher in action
The control loop for AI work — catch it before it runs, prove it after.
Get started