Adding usage telemetry to the LLM adapter broke every call site silently
What happened: to track token spend, callchat and callembed were changed to return
3 docs tagged with "infrastructure"
View All TagsConsolidating the vector store into Postgres removed a second datastore
Summary: V1 ran Qdrant as its own container alongside the Postgres already used for
One gateway URL, two credentials — chat and embeddings needed separate API keys
What happened: both SDK clients pointed at the same PLATFORMAIBASEURL and shared one