Private AI Deal Team
Five specialized AI roles take the work from financial-record organization through forensic review, diligence investigation, buyer-economic treatment, and fixed recast construction. Your team reviews the finished result in the Results & Review Desk, where every important number can be followed back through the decision, investigation, and source evidence behind it.
The deal team does the work. Your team reviews the result.
The roles
Takes every record — PDFs, spreadsheets, tax returns, statements, schedules, scans, exports, inconsistent labels, multiple entities — reads it all, identifies what exists, organizes entities, accounts, and periods, reconciles overlapping evidence, flags gaps and unreadable pages, preserves every value's source reference, and produces a complete, organized, source-traceable financial workbook that any accountant, analyst, or AI employee can work from. Also available on its own →
Produces: the organized, source-traceable financial workbook. Never: invents a value the source doesn't support.
Establishes what is defensible to analyze: which entities belong in scope, whether the records support standalone or consolidated views, which periods are supportable, and where the fiscal boundaries fall. It maps the full account-by-period record and identifies, with source citations, exactly what deserves investigation.
Produces: evidence-anchored questions for investigation. Never: claims a period the records don't support — it tells you exactly what information would be needed to support it.
Investigates each question against the evidence: support, contradiction, recurring or not, related parties, source conflicts, the quantifiable amount, missing support. Records a supported finding for every question.
Produces: established findings, with supporting evidence, counterevidence, amounts, and anything still missing named. Never: decides the buyer's economic treatment.
Decides what a buyer should economically do with established facts: approve, reject, or modify each candidate adjustment; recurring versus non-recurring; replacement compensation; market rent. Where public information is needed, it uses sanitized public research — queries that carry no confidential data.
Produces: recorded economic decisions for each supported adjustment or normalization issue. Never: performs the accounting arithmetic.
Builds reported → adjusted financials using fixed, repeatable calculations: EBITDA and adjusted EBITDA, standalone and consolidated views, eliminations, and reconciliations — with every important figure traceable to its supporting record and recorded decision.
Produces: the versioned financial recast. Never: exercises judgment.
Recast Engine
The engine builds; it does not decide. Every adjustment it applies comes from a recorded economic decision by the QoE Analyst or an authorized reviewer. The engine records which decision and which result version produced every number.
Reported → adjusted
Every step from the organized workbook to the adjusted financials is preserved and inspectable.
EBITDA
With each adjustment listed, sourced, and versioned — nothing applied without a recorded economic decision.
Views
Entity views with eliminations and reconciliations that tie.
Traceability
The economic decision, the investigation, the question, and the source page that produced it.
Results & Review Desk
Illustration — fictional labels, no figures
Unresolved items are never buried. No adjustment is applied while an issue remains unresolved — shown as $0 applied — and the exact question stays visible: your review when the judgment belongs with you, or evidence required when a missing document would resolve it. The rest of the deal team continues everything it can complete independently.
When you answer the question, override a decision, or add evidence, the affected work produces a new version you can compare with the one before it.
The reviewer sits above the system — always.
Public research
Where a workflow needs outside information — such as market compensation for a role or market rent for a property type — the system can issue sanitized research queries that contain no confidential deal data. Your deal records and internal figures are processed within your dedicated private system.
See Privacy & Architecture → for how that boundary works.
What it won't do
Next step
Show us how your team handles financial diligence today — the records you receive, the review your analysts perform, where time gets consumed, and where judgment belongs with your team. We'll map what a private AI deal team could take ownership of, where its dedicated system should live, and what it would take to put it to work on your deals.
Book an AI Workforce Consultation