AI assist

Questions are routed by a small fast model, answered by the main model, and grounded in the stored run data through vector retrieval. Agent mode lets it re-run the optimiser when you ask.

Ask anything about the latest pricing run.

Try asking

How it answers

  1. 1. A small fast model tags the topic of your question.
  2. 2. The question is embedded and matched against stored run summaries.
  3. 3. The main model answers using only those retrieved facts.
  4. 4. In agent mode it can also trigger a fresh optimisation run.