Investigate and address performance issues with the Norma chat agent.
Areas to look into:
- Response latency (time to first token, total response time)
- System prompt size and its impact on token usage
- Streaming performance and buffering behaviour
- Context assembly overhead (framework knowledge, document summaries, reporting evidence)
- LiteLLM routing and provider-specific optimisations
Investigate and address performance issues with the Norma chat agent.
Areas to look into: