AI System for Comparing Quotation Revisions

The problem

When previous and current quotations, appendices and revision notes are held in different sources, their fields must be matched manually before a technical and commercial difference summary can be prepared.

The quotation manager must compare every suggestion in the technical and commercial difference summary with the source record; missing or conflicting evidence must not replace the final decision.

If a formatting change that appears to be a scope difference is not referred for separate review, a critical difference may be missed; the record must not proceed without a human decision.

What we set out to improve

• Time taken for the technical and commercial difference summary draft to reach the quotation manager for review with its supporting evidence
• Outcomes of records referred for human review because a formatting change appears to be a scope difference
• Number of suggestions corrected after identifying the risk of missing a critical difference

How the system works

1. 01 Retrieve the previous and current quotations, appendices and revision notes from authorised sources using the record identifier
2. 02 Link source, date and version details to the shared work record
3. 03 Validate fields and units
4. 04 Prepare the technical and commercial difference summary with its supporting evidence
5. 05 Check for the exception: a formatting change appears to be a scope difference
6. 06 The quotation manager accepts, amends or rejects the draft
7. 07 Write the approved decision and its source to the audit log

Methods we used

• A source-referenced shared data schema for the previous and current quotations, appendices and revision notes
• A decision structure that separates deterministic rules from AI suggestions
• A review screen showing the technical and commercial difference summary alongside the underlying evidence
• A mandatory human review queue when a formatting change appears to be a scope difference
• An event identifier that prevents duplicate processing

Where people stay involved

Data and security

Only the fields required for this decision should be processed from the previous and current quotations, appendices and revision notes. Access to source systems should be authorised separately from the quotation manager role; personal, commercial and technical secrets should not be passed into the model context unnecessarily. Reads, suggestions and human decisions should be logged separately.

Who this suits

Frequently asked questions