Journal Yapp
Notes on operating self-hosted AI.
Notes on provider routing, team keys, usage metering, spend budgets, session capture, and useful agent context.
Why a self-hosted LLM gateway belongs between agents and model providers
Provider keys are infrastructure credentials, not application settings. The gateway keeps them in one place and gives teams a stable control plane for model traffic.
2 min readSpend quotas are more useful than request limits
A thousand tiny requests and one huge context request are not the same operational event. Budget controls need to follow cost, not just count.
2 min readAgent memory should come from real sessions
The best memory is operational evidence: what the agent tried, what failed, what fixed it, and where the next agent should look first.
2 min read