Diligence for software and SaaS acquisitions
Pactlab reconciles the revenue, reads the code history and the contracts, and prices what it finds. Every number traces back to its source. Every conclusion is signed off by a person.
AI diligence. Human instinct.
Each step keeps a link to the one before it, so a term in the purchase agreement can be traced back to the record that justified it.
Billing exports, repositories, delivery data and documents, normalized and hashed.
A drafted issue with citations, waiting for a reviewer.
An explicit, versioned input — never a sentence buried in a memo.
ARR-multiple and DCF runs from stored inputs.
Price adjustments, escrows and indemnities tied to accepted risks.
The first hundred days, linked to what diligence found.
Four flows, end to end. No mockups of mockups — this is the shape of the real thing.
Connect GitHub, GitLab, or Bitbucket. Pactlab maps contribution patterns into stars, core, supporting — and the flight risks that reprice the deal.
Animation: connecting a version-control tool, then contributors grouping into stars, core, supporting, and flight risk
Billing exports become a normalized ledger: MRR you can trace, cohorts you can point at, and management's number reconciled line by line.
Animation: billing data normalizing into a ledger, then an MRR chart and cohort retention grid
Approve a finding and watch the valuation bridge reprice. Dependent scenarios are flagged stale — never silently updated.
Animation: a finding being approved, then a valuation bridge adjusting the price downward
Questions answered with exact-page citations. No citation, no accepted finding.
Animation: a document being indexed, a question typed, and an answer appearing with a page citation
Sixteen capabilities across four arcs — one evidence loop underneath them all.
Diligence
Eight evidence streams, one normalized record. Every finding below points at its source.
Stage, target, team, permissions, tasks, and evidence status in one thin shell. It syncs to your CRM instead of trying to replace it.
MRR, ARR, NRR, GRR, churn, cohorts, and customer concentration computed from billing records — CSV or Stripe. Management's ARR is reconciled against transactions, and a reviewer signs off on the difference.
Read-only connections to GitHub, GitLab, and Bitbucket. Commit history, code-health indicators, SBOM and license findings — every finding points at the exact commit. Source code is never stored.
Contribution patterns reveal the stars: core architects, domain experts, and single points of failure, grouped where you can see them. Flight risk and concentration feed team-level valuation adjustments.
License exposure hiding in dependencies, surfaced as priced findings with evidence — not a spreadsheet of maybes.
Security posture reviewed against the frameworks your acquirers care about, mapped to findings with evidence attached.
Key terms extracted from contracts with page-level citations. Deviations from your standard positions are flagged for counsel.
Connect your VDR and evidence flows straight into diligence. No re-uploading, no duplicate sets.
Valuation
Deterministic models from versioned assumptions. Change a risk, reprice the deal.
ARR-multiple, DCF, trading comps, and LBO models built from versioned assumptions. Sensitivity tables and a purchase-price bridge come standard.
When a finding is approved, every scenario that depends on it is marked stale until repriced. The bridge shows exactly what each risk costs.
Infrastructure spend tied to unit economics, so hosting costs show up where they belong: in the numbers.
Decision
Nothing is accepted without evidence. Nothing is priced without a reviewer.
Every risk carries severity, confidence, cost-to-fix, time-to-fix, evidence, and a suggested deal consequence: price adjustment, escrow, or indemnity.
Ask questions across the entire data room. Answers arrive with exact-page citations — and a draft without a citation can never be accepted.
Integration
Diligence does not end at signing. Findings become the plan.
Accepted risks become a 100-day plan, exported to the task system you already use.
Public acquirer, private acquirer, or take-private. EDGAR ingestion and disclosure-gap analysis for public targets; data-room discovery for private ones.
Your team signs in with the identity provider you already trust — Microsoft Entra ID, Okta, or Google Workspace. Tenant isolation is enforced in the database itself.
Pactlab uses Claude to read, compare and draft. The decisions stay with the deal team.
AI extracts, compares and drafts. It never approves a finding, sets a price or makes a legal call.
Money, KPIs and scenarios come from versioned calculations and decimal arithmetic, not model prose.
No citation that resolves, no accepted finding.
Aggregate by default. HR, compensation and performance data never reach a model.
We are working with a small number of design partners on live deals. If you buy or invest in software businesses, talk to us.