Due diligence engine you own
Test your hypothesis on your
private haystack
On-premise, air-gapped AI agents that simulate a court case to test high-stakes hypotheses against your 10,000+ private company documents. Providing full evidence traceability and human-in-the-loop control, all during a coffe break
00 · The status quo
Billion-dollar decisions still rely on
manual document review.
The needle you miss breaks the deal.
The problem
Teams of junior analysts spend weeks combing through thousands of contracts, filings, and reports looking for the one clause that kills a deal. They miss things. They know the other side might find what they didn’t. Every overlooked red flag becomes a liability that compounds after close.
3–6 months
Average M&A due diligence timeline, per deal
$2–5M
Burned on a single due diligence engagement
23%
Of post-acquisition write-downs trace to documents nobody read
Not the solution
Keyword search… seriously? AI chatbots hallucinate, their context window is too small, and there is no provable way to trust the output. The other option is wrestling legacy software with a brutal learning curve and features that don’t justify the pain.
01 · Interface
Case docket
Target company revenue exceeds $50M in FY2025
847 documents · Mar 15, 2026
No material litigation risk in the last 3 years
1203 documents · Mar 15, 2026
Supply chain has single-supplier dependency
512 documents · Mar 15, 2026
IP portfolio covers core product functionality
334 documents · Mar 15, 2026
02 · Why Lodestone
Most tools make you trade off rigor, privacy, and speed. Lodestone does not.
Adversarial reasoning
Red Agent attacks, Blue Agent defends, an impartial Judge rules. Each iterative round refines the verdict, asymptotically converging on the provably correct answer.
Runs on your perimeter or don’t bother
Air-gapped, on-premise deployment. NDA-protected and classified documents never leave your network. Not even for inference. Need cloud speed or a hybrid mix? Same engine, your call.
Limitless scale, instant verdict
No cap on corpus size. 1k documents, no problem. Parallel agent execution scans thousands of pages and delivers court-grade analysis in seconds, not months.
03 · How it works
Pipeline
Ingest
Drop in contracts, filings, or reports. Lodestone parses, chunks, and embeds every document.
Hypothesize
State a claim about your documents and let Lodestone put it on trial against the full corpus.
Trial
Red Agent attacks the hypothesis while a Blue Agent defend it across multiple adversarial rounds.
Verdict
An Judge weighs both sides and delivers a structured ruling: confidence score, cited evidence, and dissenting arguments.
Ask
Ask a question or define a search target across your entire document corpus.
Retrieve
Lodestone pulls the exact passages that matter, ranked and ready to inspect.
Chat
Continue the conversation and get insights. Every response is grounded with inline citations you can verify.
Features
White-box traceability
An onion-like audit trail: peel from verdict to arguments to evidence to source citations. No black boxes.
Verdict in seconds
Parallel agent execution delivers court-grade analysis at the speed of a coffee break.
Human-in-the-loop
Override agent arguments, inject domain expertise, and steer the trial at any point.
Massive corpus, no limits
Ingest tens of thousands of documents. No cap on corpus size. Scales linearly with your data.
Chat with docs
Search the corpus, inspect the right passages, then keep digging in chat. Every answer stays grounded with inline citations.
Adversarial court simulation
Red Agent attacks, Blue Agent defends, an impartial Judge rules. Multiple rounds ensure no claim goes unchallenged.
Asymptotic certainty
Each adversarial round mathematically narrows the error margin. More iterations, higher confidence. Provably converging.
On-premise & air-gapped
Built for NDA-bound diligence and classified environments. Your data never leaves your perimeter.
Data ownership
Choose where Lodestone runs. Every deployment offers the same features.
Air-gapped
Complete network isolation. Everything runs offline with local models.
- Local models only
- Classified environments
- Document secrecy
Hybrid
The middle ground. Run the core pipeline on your infrastructure while offloading specific tasks.
- Granular control
- Mix local & remote models
Cloud
Managed deployment on our infrastructure. We handle uptime, scaling, and updates.
- Managed infrastructure
- Auto scaling
- Low maintenance
Submit a hypothesis, let adversarial agents run the trial against your private documents, and get a traceable verdict. All during a coffee break.