Chat with your project
Ask questions about your Brief + Artifacts.
After a run completes, you can chat with the project. The chat has context for your Brief, your artifact bundle, and the graph of relationships between them.
What you can ask
- "Why is this requirement here?" — the chat walks you through the NGO → requirement derivation, citing your Brief
- "Which components depend on the comms subsystem?" — graph queries over the decomposition
- "Summarize the top 3 risks" — synthesis over the risk register
- "What's missing from the design?" — gap analysis using the phase-gate criteria
- "Compare this trade study to one I'd do for [different approach]" — reasoning over alternatives
What it knows
The chat has access to:
- Your full Brief
- The full artifact bundle (every node, every citation)
- The graph of relationships (NGOs ↔ requirements ↔ components, etc.)
- The render in Markdown
It does not have access to:
- Other projects in your account
- Anything outside the Brief and artifacts
- General-knowledge facts beyond its training data (so don't ask "what's the latest in laser propulsion?" — ask domain questions about your project)
How to use it
Open the chat panel from any page within a run (status, complete, explore). Type a question. Streaming response with citations into your artifact bundle.
Cost
Chat consumes CU. Each chat message is approximately 0.5–2 CU depending on how much context the model needs to load. The current chat session's CU is shown in the panel.
NOTE
Chat is a research and review tool, not a design tool. If you want
to change an artifact, use [Revisions](/docs/features/revisions) so
the change is structured + audited. Chat doesn't modify your bundle.
Limitations
- The chat can be wrong, especially on judgment calls. It's working from the same artifact bundle you can read directly
- It will not refuse to answer — if you ask a leading question, it will give you a leading answer. Frame questions neutrally
- For ITAR/EAR concerns: don't paste sensitive content into the chat that wasn't already in your Brief. See Acceptable Use Policy