From €1,500, in 1–3 weeks. An LLM feature with a safety judgement, output checks and tests.

I add a working LLM feature to your app and build it so it can be trusted: the user's own key, a safety judgement before the reply, a check on the draft, tests, and no user text in logs. What's included - The feature itself, wired into your app (chat, writing help or similar), using the customer's own Claude or OpenAI key. - A safety step before the reply and a check on the draft after it. - Encrypted local storage where it fits, and logs that hold no user text. - Tests that run without a paid model, using a scripted mock. - Short documentation of how it works and how to change it. What I need from you: access to the codebase, a description of what the assistant should do, and your target platform. A short call at the start to agree the scope. Delivery: 1-3 weeks, depending on the app. Fixed price agreed before I start.

Good to know: the scope is agreed in writing first. Changes after that are quoted separately. I work mainly in Go, Python and JavaScript.