Human review in financial diligence

AI-assisted diligence brings several different activities into the same conversation: document extraction, deterministic calculation, anomaly identification, interpretation, and professional approval. Treating them as one undifferentiated output obscures their different risks.

Extraction asks what a source says. A scanned statement or irregular table may require a different method from a structured accounting export. The extracted value still needs a source location and, where appropriate, verification.

Calculation asks how a value is derived. Financial figures should follow defined logic using identifiable inputs. A fluent explanation does not establish that the arithmetic or account mapping is correct.

Interpretation asks what the evidence means for the business and transaction. An unusual item, customer concentration, or margin movement may deserve attention without supporting a definitive adverse conclusion.

Approval is a professional decision. The reviewer must consider scope, evidence quality, accounting context, and the intended use of the deliverable before accepting a conclusion.

The practical rule is that software may produce any figure it can trace, and may propose anything it can explain, but a figure without a source is not reported and a proposal without a reviewer is not accepted.

Addback uses deterministic application code for the figures that drive its screening checks. Unsupported evidence is marked not assessable rather than filled with a model-generated estimate.

Where Addback uses its optional AI extraction layer, those outputs are kept separate, carry an AI origin and verification requirement, and do not replace authoritative deterministic figures.

The same boundary applies to narrative. Generated text that summarizes a schedule is a convenience; it is not evidence, and it should be labeled so a reader never mistakes a summary for a source.

Review should be directed at material questions, not reduced to a final checkbox. Give reviewers the source, calculation, rationale, uncertainty, and proposed next action needed to challenge a finding.

What a reviewer needs on screen

• The finding in one sentence, with the amount and the period.
• The rows or documents that produced it, one click away.
• The method used and how confident it is, stated plainly.
• The options: accept, reject, hold for evidence, or ask management.
• The effect of each option on the adjusted figures.

Keep disagreement visible. A reviewer may reject a candidate adjustment or require additional evidence. Preserving that decision is more informative than silently changing the final number.

Order the queue by materiality. A reviewer who sees fifty immaterial flags before the one that moves price will stop reading before they reach it.

Finsider offers different levels of support: Addback for the first screen, Platform for work performed by your team, and Advisory for a scoped CPA-led engagement. The professional responsibilities should be clear in each workflow.

In a self-serve workflow, the firm or deal team is the reviewer and the deliverable is their work product. In an Advisory engagement, a CPA reviews and signs, and the scope letter defines what that signature covers.

This Finsider Labs note describes a design and review framework. It does not claim that AI eliminates errors, replaces a CPA, or guarantees an investment outcome.