When an AI response is especially large or complex, Tesseract automatically extracts it into an artifact — a dedicated view separate from the chat flow. This keeps conversations readable while preserving the full output.
A response is extracted as an artifact when it exceeds any of these thresholds:
| Metric | Threshold |
|---|---|
| Total blocks | More than 12 blocks |
| Total characters | More than 8,000 characters |
| Table rows | More than 30 rows |
| Chart data points | More than 50 data points |
| Code lines | More than 200 lines |
These thresholds are configurable by the system administrator.
When a response is turned into an artifact, you'll see an artifact link in the chat instead of the full content. Click the link to open the artifact in its dedicated view.
<aside> 💡
Tip: If you'd prefer shorter, in-chat responses, try asking more specific questions. Broad prompts like "Explain everything about machine learning" are more likely to produce artifacts, while targeted prompts like "What is gradient descent?" will stay in the chat.
</aside>