Due diligence that reads the whole picture.
A $200M deal generates 2,000 documents across legal, financial, technical, and regulatory workstreams. Today those workstreams run sequentially, each team blind to what the others are finding. Purple8 ingests the entire data room, builds a relational knowledge graph of the target, and runs all five workstreams in parallel, so the findings talk to each other from day one.
The current state
A $200M deal costs $2–7M to review. Most of that is information processing.
Current advisory spend per deal
With Purple8
Advisory fees don't disappear. Experienced judgement has real value. What changes is what that time is spent on.
How it works
Seven phases, one platform, all findings connected.
Data room ingestion
DocIntelEvery document in the virtual data room (contracts, financials, filings, technical specs, correspondence) is ingested via DocIntel. Entities, obligations, counterparties, dates, and relationships are extracted automatically and written into the knowledge graph. A data room that would take a paralegal team two weeks to index is processed in hours.
Knowledge graph construction
graph.*The target company becomes a living graph. Contract obligations become edges. Counterparty relationships, IP dependencies, regulatory filings, and related-party transactions all have typed representations in the structure. Cross-references that require reading 40 contracts simultaneously emerge from a single graph query.
Parallel specialist review
Multi-agent swarmFive agents run simultaneously, each writing findings to the shared graph as they work. Legal flags change-of-control triggers, non-standard indemnities, and IP ownership gaps. Finance identifies customer concentration, recurring vs one-off revenue, and covenant exposure. Technical maps single-engineer dependencies and licence conflicts. HR flags key-person risk and compensation anomalies. Regulatory maps pending matters and jurisdiction exposure. None of them wait for the others.
Stakeholder & sentiment mapping
Opinion EngineThe opinion engine traverses the graph to model which executives are likely to stay, which customers are at risk, and which investors may push back on the deal. Document signals (board minutes, investor communications, employment agreements, customer correspondence) are weighted by recency and source credibility. The output is a stakeholder position map before the first management presentation.
Scenario stress-testing
Scenario SimulationThe live deal graph is cloned into isolated simulation namespaces. Each scenario applies a specific shock: the two most graph-central engineers leave, the anchor customer churns, the regulatory change pending in Brussels passes. Graph algorithms score the impact on each scenario. The advisory team reviews the stress-test results before deciding which risks require price adjustment, escrow, or earnout protection.
Deal structure optimisation
quantum.optimizeEarnout schedule, escrow amount, and retention package allocation across key employees are binary allocation problems under competing constraints. The quantum optimisation layer formulates them as QUBO problems and finds the Pareto-optimal deal structure: one that protects the acquirer's downside while remaining acceptable to the target's board.
Staged approval and audit trail
Journey EngineThe deal moves through defined approval gates: IC memo, legal sign-off, board approval, regulatory filing. Each gate is a mandatory human decision recorded as an immutable graph edge. When a regulator or a court asks to see the decision trail, the full chain is in the graph, traversable and exportable in minutes, showing who reviewed what based on which findings and when.
Why graph matters here
The risks that matter are between documents, not inside them.
A change-of-control clause in a supplier agreement only becomes critical when you know that supplier accounts for 40% of gross margin. That fact lives in a different document in a different workstream.
An IP ownership gap in a patent assignment only becomes a deal-breaker when you know the product depends on that specific patent. That context is in the technical architecture documentation that the legal team never reads.
Purple8 stores every entity from every document in the same graph. Cross-workstream risks surface automatically because the graph traversal doesn't respect the boundaries between teams.
CONTRACT → SUPPLIERChange-of-control consent required
SUPPLIER → REVENUE43% of gross margin dependency
PATENT → PRODUCTCore feature relies on this IP
PATENT → ASSIGNEEAssignment incomplete: founder holds it, not the company
ENGINEER → CODEBASESole author of payment module
ENGINEER → VESTINGCliff in 3 months, flight risk
Each row is a graph edge. In a traditional review, these six facts live in six different documents reviewed by three different teams.
Who uses this
Built for the firms that run the deals
Boutique M&A advisory firms
Run more mandates with the same team. A four-person firm that currently handles 8 deals a year can handle 25. The differentiation in pitch: faster, more thorough, and you can show the client the knowledge graph of their own target.
Big 4 Transaction Services
Replace the associate-hours spent on document indexing and first-pass review. Senior staff spend time on findings that require judgement. The audit trail satisfies the same documentation standards as the current process.
M&A law firms
Contract review that surfaces every change-of-control clause, every indemnity anomaly, every IP gap across a 2,000-document data room in hours. Partners review the graph, not the stack of paper.
Corporate development teams
Run your own preliminary due diligence before engaging external advisors. Arrive at the first advisor meeting having already mapped the target's risk landscape. Negotiate the advisory mandate from a position of knowledge.
Private equity firms
Build a consistent due diligence knowledge graph across every portfolio company. When a second acquisition in the same sector comes along, you start from an existing graph rather than from scratch.
FAQ
Questions we get from advisory teams
What does Purple8 actually do in M&A due diligence?
Purple8 ingests the entire data room via DocIntel (contracts, financials, cap tables, IP filings, employment agreements, regulatory correspondence, technical documentation) and builds a relational knowledge graph of the target company. A multi-agent swarm then runs legal, financial, technical, HR, and regulatory review simultaneously — each agent writing findings as typed graph edges. The result is a cross-linked knowledge structure that reveals relationships and risks that siloed document review misses entirely.
How does this replace the current advisory process?
It doesn't replace the advisory firm — it changes what their team spends time on. The agents handle document ingestion, entity extraction, clause identification, cross-reference detection, and first-pass risk flagging. The advisors spend their time on the findings that matter: the edge cases, the negotiation strategy, the client relationship. A team that currently runs 5 deals a year can run 20 with the same headcount.
How does the audit trail work for regulatory purposes?
Every finding, every agent action, every human approval decision is written as an immutable graph edge with agent identity, timestamp, and provenance back to the source document. When a regulator asks to see the decision trail, the entire chain is in the graph, traversable and exportable. Nothing is reconstructed from memory or emails after the fact.
Can change-of-control clauses be detected automatically?
Yes. DocIntel extracts contract obligations and the legal agent traverses the graph to find all agreements with change-of-control triggers, consent requirements, or acceleration clauses. Because they are stored as typed edges in the knowledge graph — not buried in document text — they can be queried across the entire contract portfolio simultaneously. The agent surfaces every affected agreement in one pass.
What does scenario simulation add to due diligence?
After the knowledge graph is built, you can clone it into isolated simulation namespaces and apply shocks: remove the two engineers with the most graph centrality (key-person risk), remove the top customer (revenue concentration), apply the pending regulatory change in the target's primary market. Graph algorithms score the impact on each scenario. The quantum optimisation layer then finds the optimal deal structure (earnout schedule, escrow, retention packages) under the constraints revealed by the analysis.
Capabilities used in this workflow
GraphRAG →
Query across the entire deal graph
Multi-Agent Systems →
Five workstreams running simultaneously
Opinion Engine →
Stakeholder position mapping
Scenario Simulation →
Key-person, customer, and regulatory stress tests
Quantum Optimisation →
Earnout and escrow structure optimisation
Workflow Orchestration →
IC, board, and regulatory approval gates
Run your next deal on a knowledge graph
We work with advisory teams to map their current process onto Purple8. The first conversation is about your deal flow, not a product demo.