Thinking Traces make the AI's reasoning visible. You see the process, not just the result.
What They Are
When the Brainstorm Reactor uses a heavy reasoning model, the AI generates an internal chain-of-thought before producing its final response. This reasoning is captured and displayed as a Thinking Trace — a collapsible section above the response.
Visual Behavior
| State | Display |
|---|---|
| During generation | Animated "Thinking…" indicator with elapsed time counter |
| After generation | Collapsed accordion with token count badge |
| Expanded | Full chain-of-thought text in a visually distinct panel |
Click the Thinking Trace header to expand or collapse it.
What You See Inside
The trace contains the model's raw reasoning: hypotheses it considered, alternatives it rejected, logical steps it followed, and trade-offs it evaluated. This is the unfiltered cognitive process — often longer and more nuanced than the final output.
Why They Matter
Standard AI interfaces present outputs as finished artifacts. This creates Automation Bias — the tendency to accept AI conclusions uncritically because the reasoning is hidden.
Thinking Traces counter this by making the reasoning auditable. You can verify whether the AI actually considered your constraints, whether it evaluated counterarguments, or whether it took shortcuts.
Which Models Support Them
Thinking Traces only appear when the active model supports chain-of-thought output. Models without this capability (standard chat models) produce responses directly, without a trace.
The active model is configured in the System Config panel under "Text Model."