We build the layer between a company and the AI tools its people use. Most companies I talk to either ban ChatGPT and watch people use it anyway, or allow it and have no idea what goes in. We're trying to give them a third option: rules instead of bans.
What exists today, all of it early:
- AI Usage Control: a managed browser extension plus an admin console. It checks a prompt before it's sent to ChatGPT and lets it through, warns, masks or blocks it. Runs on the company's own servers.
- shim gateway: an OpenAI / Anthropic / Gemini compatible gateway that masks personal data before the request leaves and puts it back in the response. Open source core.
- shim-cli: the same idea for coding agents, as a local hook. Open source, no account, no telemetry.
Turkish data is our home turf: TCKN, VKN and TR IBANs with checksums, not just regex.
Happy to talk about any of it
getshim.tech github.com/DurthVadr mertcan@getshim.tech