Long answers finish instead of being cut off
AI Agents, the Assistant and every generator now let the model write until the answer is complete.
What this solves
Some long deliverables stopped part-way and asked you to try again: a long article, a website audit with many pages, a large ICP or product catalogue, a chat answer with several sections. Each retry re-sent the whole request, so a cut-off answer cost more than a finished one. Separately, a few generators failed with a timeout on slow answers rather than waiting for them.
What changed
- AI Agents, the Assistant, Knowledge Centre generators, Website Audit, Competitors, Sales and Campaign tools no longer impose their own output length limit. The model writes until it is done, up to its own maximum.
- Generators wait for a slow answer as long as text keeps arriving, instead of giving up at a fixed number of seconds. A stalled answer still fails cleanly.
- Deliberately short outputs — labels, one-line summaries, quick classifications — are unchanged.
- A run that does reach the model's absolute limit is now recorded like any other, so it shows up in usage history.
Where to find it
Nothing to configure. Run any agent, ask the Assistant a long question, or run a Website Audit as before.
Useful for
Anyone producing long-form content or large structured documents — articles, audits, personas, product catalogues, multi-section briefings — who has seen "try again" or a timeout on a long run.
About Checkgrow
Checkgrow is the growth platform where AI Agents write, research and plan from your own company knowledge. Every deliverable — from a LinkedIn post to a full website audit — is grounded in what your business actually offers, so the output is ready to use rather than ready to rewrite. See how it works at