Skip to content

Reviewing Answers

The review workflow is the core of Tendrex. After the AI generates draft answers, your team steps through each one in a dedicated full-screen review interface to verify, edit, and approve.

The review page provides a focused, full-screen environment:

  • Split panel layout - answer on one side, source documents on the other (resizable)
  • Previous/Next navigation - step through questions sequentially
  • Keyboard shortcuts - navigate and take actions without leaving the keyboard
  • Auto-advance - after taking an action, automatically moves to the next pending answer
  • Onboarding guide - first-time users see a walkthrough of the review interface

For each question-answer pair, reviewers can take the following actions:

ActionDescription
ApproveAccept the answer as-is. It is marked as final and added to the answer library.
EditModify the answer text, then approve the edited version.
RejectDiscard the AI answer with a comment explaining why. The feedback is stored and used to improve future generations.
ProvideManually provide an answer without AI assistance.
AssignRoute the question to a specific team member or team for specialist input. Sends an email notification if configured.
Not RequiredMark the question as not applicable to this tender.

Each AI-generated answer includes a confidence indicator:

  • High - the AI found strong, relevant source material and the answer closely matches existing approved answers.
  • Medium - some relevant material was found but the answer may need verification.
  • Low - limited source material was available. The answer should be carefully reviewed.
  • N/A - no relevant source material was found. The answer was generated from scratch.

Every AI answer includes citations pointing to the specific passages in your source documents that were used. Click a citation to:

  • View the source document in the built-in viewer (PDF or Word).
  • See the highlighted excerpt that the AI referenced.
  • Check the page number and document name.

This gives your team full traceability over every answer.

Every change to an answer is tracked with full version history:

  • View all previous versions of an answer
  • See a diff view showing exactly what changed between versions
  • Track who made each change and when

This provides a complete audit trail from the initial AI draft through to the final approved answer.

Each answer has a per-answer activity trail showing:

  • When the answer was generated
  • Review actions taken (approve, reject, edit, assign)
  • Who performed each action and when
  • Comments and rejection reasons

Before approving, you can use AI to transform any answer. Six actions are available:

ActionWhat it does
ImproveFix grammar, improve clarity and professionalism without changing facts
ExpandAdd details, examples, and metrics - roughly doubles the length
CondenseReduce to key points - roughly halves the length
FormalRewrite in formal UK public sector tone
Bullet PointsConvert to scannable structure with headers and bullets
StrengthenAdd compelling evidence, specific commitments, and measurable outcomes

AI edits are applied as a new version, so you can always revert.

Each answer supports threaded comments for team discussion:

  • Add comments with text highlighting to reference specific parts of an answer.
  • Thread replies to keep conversations organised.
  • Resolve comments when the feedback has been addressed.
  • Comments are visible in the activity trail and audit log.

When you approve an answer, it is added to your answer library. You can further promote it to canonical status. A canonical answer becomes the preferred response for similar questions in future tenders. The AI will prioritise canonical answers when generating drafts, improving consistency across submissions.

For efficiency, you can select multiple questions and apply bulk actions:

  • Approve all selected
  • Reject all selected
  • Assign all selected to a team
  • Mark all as not required

The tender dashboard shows an overview of review progress - how many questions are approved, pending, rejected, or assigned. Use this to track completion and identify bottlenecks.