n8n human in the loop: connect a specific approval to a specific post

Make the proposed post visible before the posting step

The 13 June 2025 tutorial uses Telegram to review AI-generated tweets before they go live. It connects an n8n content workflow with human feedback, aiming to improve posts that otherwise feel generic. A human checkpoint is useful when the reviewer sees the actual content and destination, not just a summary that a post is ready.

1. Follow the concept at 0:12 and Telegram workflow at 0:41.
2. Choose one test draft and keep its full text, intended account and version identifiable.
3. Inspect what reaches the Telegram reviewer and confirm it contains enough context to approve or request a concrete change.

Route feedback to a revised draft, not directly to publication

Feedback routing appears at 3:00, chatbot testing at 4:00 and real-time edits at 5:33. These stages need to preserve the relationship between the reviewer’s response and the content it describes. A request to improve wording should produce another reviewable draft, while a rejection should not be interpreted as an instruction to post.

1. Test a specific revision request and compare the new draft with both the feedback and the original intended message.
2. Check a rejection or unresolved response and inspect whether the workflow actually remains held.
3. Make the revised version visible again before considering it approved; approval of an earlier draft should not silently carry over to changed text.
4. Keep unrelated messages distinguishable from an explicit decision about the current draft.

Inspect the final action and its receipt

The final posting workflow appears at 6:43. The description does not provide exact node settings or approval-message syntax, so use the recording for the demonstrated connection. Its alternate titles promise reliability and scale, but a review loop does not independently guarantee factual accuracy, correct account mapping or fault-free delivery.

1. Verify that the posting step receives the approved text and intended destination rather than a previous queued version.
2. Inspect the actual post or delivery result and keep it linked to the reviewed draft.
3. Check uncertain outcomes before retrying so an approval workflow does not create duplicates.
4. Review the resulting text for specific facts and natural wording; human approval is not a substitute for inspecting unsupported claims.

The takeaway

An effective approval loop preserves exactly what was reviewed, what feedback changed and what was ultimately posted.