Refining your analysis
Re-running analysis, responding to clarification requests, and getting better results.
Sometimes the first analysis isn't the last. Documents get updated, you realize you forgot to set up criteria, or you want to dig deeper into something specific.
When to re-run analysis
Good reasons to re-run:
- Vendor submitted updated documents - they fixed issues you flagged, or sent additional materials
- You re-extracted or changed your RFP requirements - criteria and lots are projections of your approved requirements, so re-extracting and approving them, then re-syncing criteria and lots, gives the analysis a fresh structure to work against
- You added or changed evaluation criteria - new criteria need a fresh analysis to extract values
- You added lots - the analysis needs to re-examine bids through the lot structure
- Documents were re-uploaded - maybe you had the wrong version of a specification
Not-so-good reasons:
- You don't like the results - re-running the same documents won't change the findings substantially. AI analysis is largely deterministic for the same input.
- One finding seems wrong - check the evidence first. If the AI genuinely misread something, a re-run might produce the same mistake. Contact support if you think there's a systematic issue.
How to re-run
Open the bid under Bids in the sidebar and click Re-check on its latest check. A new analysis starts from scratch using the current documents and criteria. The previous results stay available for reference.
Each re-run costs additional credits based on the document size and complexity. It's the same cost as the original analysis.
Responding to clarification requests
Occasionally, the AI needs your help during analysis. Maybe a document references an appendix that wasn't uploaded, or there's ambiguous language the AI can't resolve on its own.
When this happens:
- The analysis pauses automatically
- You'll see a question displayed on the results page
- Type your answer and submit
- The analysis resumes, incorporating your response
Check back periodically if you're not watching the analysis live. A clarification request just sits there waiting until you respond.
Batch analysis
If you have multiple bids to analyze (or re-analyze), use Analyze all on the Bids page. This queues up every bid and runs them one after another.
Batch analysis is useful when:
- You've just added 5 new bids and want to analyze them all
- You changed criteria and need to re-run everything
- You have a deadline and want all analyses done without babysitting each one
The analyses run sequentially (one finishes, the next starts) and you can see progress on each.
Asking the AI Assistant
To ask about the documents or the analysis results in plain language, open AI Assistant in the top bar and pick this procurement as the working context. It searches the RFP and every bid at once and reads completed analyses, so it can answer questions such as:
- "What does Vendor A say about their delivery timeline?"
- "How do Vendor A and Vendor B compare on team experience?"
- "What evidence supports the 'missing financial guarantee' finding?"
Answers name the file and the section or page they came from. See AI Assistant.
Getting better results
A few things that noticeably improve analysis quality:
Upload everything. If the RFP references appendices, forms, or standards - upload them under Tender documents → Documents. The AI can only check requirements it knows about.
Set up criteria first. Configure criteria on the Evaluation criteria tab under Tender documents before running analysis. Analysis with criteria is more focused and produces richer results than a general compliance check.
Clean up your documents. If a 300-page proposal includes 100 pages of standard marketing material, that's noise. The AI has to process it, and it costs you credits. If possible, upload only the relevant parts of very long documents.
Use clear file names. "Technical_Proposal_CompanyX.pdf" is better than "Document_v3_final_FINAL.pdf". The AI sees file names and uses them for context.
Each analysis run is independent. Re-running doesn't build on previous results - it starts fresh. This means you always get a clean assessment of the current state of the documents.