Conflicting numbers
Revenue, retention, EBITDA and customer metrics often do not reconcile across source files.
AI diligence agents for private equity
Know what could break the deal before you buy the company.
Keystone verifies management claims, investigates discrepancies and surfaces the financial, customer and contract risks that could change your investment decision.
Built for private equity deal teams conducting high-stakes diligence.
Management Claim Verification
Revenue Quality
Customer Retention Analysis
Customer Concentration
EBITDA Add-Back Review
Contract Diligence
Forecast Challenge
Management Q&A
Source-Linked Evidence
Investment Committee Preparation
Management Claim Verification
Revenue Quality
Customer Retention Analysis
Customer Concentration
EBITDA Add-Back Review
Contract Diligence
Forecast Challenge
Management Q&A
Source-Linked Evidence
Investment Committee Preparation
The problem
Private equity teams have the data. They do not always have the truth.
During diligence, deal teams receive CIMs, financial statements, customer files, contracts and thousands of data-room documents. Keystone helps teams verify what is real, identify what does not add up and focus diligence on the issues that could change the deal.
Unsupported claims
Management narratives can sound convincing before the underlying evidence has been tested.
Manual investigation
Associates repeatedly clean files, compare versions, reconcile totals and draft questions.
Disconnected conclusions
Findings are not always clearly connected to valuation, leverage, forecasts or returns.
Product example
See what Keystone actually does.
A professional diligence surface for investigating one complete claim, not a chatbot detached from the sources.
Management claim
Customer retention is 95%
Keystone finding
Verified gross revenue retention: 84%
Potential impact
Revenue forecast may be overstated.
Management excluded partially churned customers and used a narrower customer population than the underlying revenue file.
Request management's retention calculation, included customer population and treatment of acquired, renamed and partially churned accounts.
How it works
From data room to verified conclusion.
Keystone AI agent
Running
Active inputs
Connect
CIM, financials and data room
Deal materials organized by source
Extract
Management claims
Revenue, customers, margins and growth
Test
Evidence and calculations
Conflicts traced to source records
Resolve
Open diligence issues
Tracked until reviewed by the team
Live workflow
Source-backed finding
Management claims 18% recurring revenue growth. Keystone is locating the supporting customer file.
Connect
Extract
Test
Investigate
Drive follow-up
Live outputs
Use cases
Built around the diligence work that changes the investment decision.
Revenue Quality
Reconcile reported revenue with customer-level data and identify unusual, nonrecurring or unsupported revenue.
Customer Retention
Calculate churn, gross retention, net retention and expansion using a consistent methodology.
Customer Concentration
Group related entities and uncover hidden exposure across subsidiaries, locations and inconsistent customer names.
EBITDA Verification
Challenge add-backs, identify recurring expenses and estimate a more defensible normalized EBITDA.
Contract Diligence
Extract pricing, renewal, termination and change-of-control risks from material agreements.
Forecast Challenge
Test management projections against historical performance, sales pipeline, churn and operating evidence.
Differentiation
Most AI tools organize the deal. Keystone investigates it.
Keystone is designed for closed-loop diligence. It helps determine what is true, what remains unsupported and what the deal team should do next.
Traditional finance AI
Keystone
Searches documents
Identifies management claims
Summarizes files
Tests claims against evidence
Extracts tables
Performs independent calculations
Answers one-time questions
Generates precise follow-up questions
Shows information
Shows potential investment impact
Traceability
Every conclusion leads back to the source.
Every claim, calculation and conclusion is linked to the underlying evidence so a deal team can inspect the exact document page, spreadsheet row, calculation or contract clause behind it.
Source-level citations
Transparent calculations
Document version history
Conflicting evidence side by side
Human approval status
Complete audit trail
Human control
AI speed. Investment-team control.
- Human approval before questions are sent to management.
- Human approval before material diligence issues are closed.
- Human approval before official model assumptions are changed.
- Human approval before investment committee materials are published.
Security
Designed around confidential deal information.
Enterprise security roadmap available during pilot discussions.
Encrypted data handling
Deal-level permissions
Role-based access controls
Human approval workflows
No training on customer data
Configurable deletion and retention policies
Design partners
Help build the diligence agent your team would actually use.
We are working with a limited number of private equity professionals to validate Keystone's initial workflows across customer revenue analysis, management claim verification and diligence issue tracking.
Request a demo
Find the issue before it becomes the investment.
See how Keystone can help your team verify management's claims, investigate what does not add up and focus diligence on the risks that matter.