Diligence Is Not Done Until the Source Survives the Conclusion

A diligence conclusion is only as useful as the path back to the evidence.

A deal team may review a CIM, leases, operating statements, debt documents, environmental reports, sponsor materials, public records, and the underwriting model. Summarizing those files is not the hard part. The hard part is keeping every conclusion connected to the right source and surfacing disagreement before it reaches committee.

Build the evidence spine first

For every extracted fact, preserve:

• source document and stable location;
• document date and reporting period;
• entity and property identity;
• extraction method and confidence;
• reviewer status;
• later corrections or superseding evidence.

A paragraph in an IC memo should be able to point back through that chain.

Compare against the acquisition box

An agent becomes useful when it can compare extracted facts with the firm's own criteria. That may include geography, basis, yield, occupancy, lease concentration, sponsor history, debt structure, or capital requirements.

The criteria must be explicit and versioned. A hidden prompt is not an investment policy.

Surface conflict, not consensus

The CIM, rent roll, model, and third-party report may describe the same fact differently. The agent should not manufacture one clean answer. It should show the competing values, source dates, and materiality.

Separate screening from judgment

Public-record and counterparty screening can gather evidence and route potential matches. A named reviewer still determines whether a record belongs to the party and whether the finding matters to the deal.

Deliver committee-ready work

The finished package can include an abstraction table, open-item log, exception report, draft IC memo, and source packet. Each conclusion should carry enough evidence that the reviewer can challenge it without reopening the entire data room.

That is the difference between document chat and controlled diligence.