Summary
Several major AI models (Claude, OpenAI o1/o3, Gemini) now expose their internal reasoning/thinking process as part of the streaming response. Askimo should surface this to users rather than silently discarding it.
Motivation
Users increasingly expect to see how the AI arrived at an answer, not just the final response. This is especially valuable for complex tasks - debugging, multi-step reasoning, planning - where the thinking trace helps users verify correctness and build trust in the output.
Proposed behavior
- Thinking/reasoning content is displayed in a collapsible section above the final response, e.g. "💭 Thinking..." that expands to show the full trace
- Collapsed by default to keep the UI clean; expanded on click
- Streamed in real time as tokens arrive, same as the main response
- Visually distinct from the main response (e.g. muted color, italic, or a dedicated style)
- User preference to always show / always hide thinking blocks
Notes
- If the streaming response already passes thinking tokens through, this is primarily a UI change
- Should gracefully no-op for models that do not expose reasoning
Summary
Several major AI models (Claude, OpenAI o1/o3, Gemini) now expose their internal reasoning/thinking process as part of the streaming response. Askimo should surface this to users rather than silently discarding it.
Motivation
Users increasingly expect to see how the AI arrived at an answer, not just the final response. This is especially valuable for complex tasks - debugging, multi-step reasoning, planning - where the thinking trace helps users verify correctness and build trust in the output.
Proposed behavior
Notes