Normalise before you compare
The useful AI task here is structural: take each quote and force it into the same shape — scope items, inclusions, explicit exclusions, provisional sums, assumptions, programme, payment terms.
Once three quotes sit in the same structure, the gaps announce themselves. What one priced, another excluded. What one assumed, another queried.
- A line-by-line scope matrix across all tenderers
- An exclusions list consolidated and de-duplicated
- Provisional sums flagged with what triggers them
- Programme and access assumptions stated side by side
- A question list to send back before any award