Why a subscription? AI inference and vector search aren't free. Every diagnosis pays for the language model that writes the answer, the embeddings that find the right manual pages, and the database that holds them. The subscription covers the cloud bill plus the time it takes to keep the corpus current. When open-source models get good enough to host locally, prices come down.

What's a diagnosis? One diagnosis = one chat conversation. Hit + new to start a fresh one; every follow-up inside the same thread (clarifications, “what about the EOS switch?”, “tried that, still broken”) is free and doesn't count again.

The honest assessment. This is an early-access tier. We've cleared 3 of 6 of our internal eval targets. Citations and actionability are solid; factual accuracy still varies by machine. Subscribe if that math works for you.