The Vault Standard · v1.0
Purchase agreements disclaim it. Insurers pay out on what it hid. Courts sanction the dumping tactics it enables. legalfiles.ai verifies what's actually inside the room — every file integrity-checked, every finding cited back to the seller's own documents, every unverifiable claim labeled unverifiable — before you sign.
For M&A attorneys, paralegals, diligence teams, independent sponsors, and sell-side advisors — in any industry.
of e-discovery production cost is document review — the single largest cost center in diligence-scale work.
RAND Institute for Civil Justice, “Where the Money Goes,” 2012representations & warranties insurance policies draws a claim — what diligence missed, priced by the insurers who pay for it.
AIG global RWI claims data, 2018of R&W claims now arrive more than 12 months after closing — they survived diligence and the first audit cycle.
Aon Transaction Solutions Global Claims Study, 2025of all recovery paid traces to financial-statement breaches — just 14% of claims. The biggest losses sit exactly where document integrity fails.
Aon Transaction Solutions Global Claims Study, 2023The rule set
The Vault Standard is not a feature list. It is the rule set every output of the platform is required to meet — and the reason our findings hold up when a deal team, a counterparty, or a court reads them.
Every finding carries a citation back to the seller's own documents — the file, the location, the line. A claim that cannot be traced does not ship. Not flagged as uncertain. Not footnoted as approximate. It does not ship.
If it can't be traced, it isn't said.
Every file in the room is byte-verified against its claimed type before anyone reads a word of it. Mislabeled, corrupted, and unreadable files are surfaced by name and count — not smoothed over, not silently skipped.
The container is verified before the contents are read.
When a marketed figure depends on source data that was never delivered — the native databases, the underlying ledgers, the executed versions — the figure is labeled unverifiable. It is never estimated, averaged, or “directionally” rescued.
Missing source, missing claim.
A data room is not a snapshot — documents arrive late, get replaced, and occasionally disappear. The room is re-verified as it changes, and every late addition and quiet replacement is flagged as it lands.
The room never goes stale — and never silently changes.
The UK High Court found fraud in HP's $11.1 billion acquisition of Autonomy — and found that roughly 80% of HP's loss came from its own diligence failures. The fraud was real. So was the bill for not catching it.
Under Delaware's ABRY line, anti-reliance clauses cut off fraud claims based on data-room materials. In ChyronHego v. Wight (2018), a claim over “misleading documents and projections placed in the data room” was dismissed.
Del. Ch. — opinions and firm analyses, cited in product researchCourts sanction “document dumps” — hundreds of thousands of pages produced to bury the responsive few. The tactic is documented, named, and punished in discovery; in a data room, it has no referee at all.
United States v. Quebe (2017); Arbor Realty v. Herrick Feinstein (2016); Stooksbury v. Ross (2013)What survives contract defenses is what was never delivered: the tax liability not in the room (Bertelsmann, 2021), the customer loss hidden from it (TransDigm, 2013). Absence is the hardest thing to see — so we look for it directly.
Del. Ch. — opinions as citedRunning proof, not a roadmap
WellSite — the energy-vertical deployment of the same engine — ran the Standard against a live oil & gas deal room. Not a demo dataset. A real room, marketed to real buyers, with real gaps. Every number on the right is a computed, receipted finding, cited to the seller's own files.
legalfiles.ai is that engine with the energy specifics removed and the legal workflow built in: intake and index normalization, review triage, Q&A tracking, red-flag memo assembly, and the diligence summary — the five places your team's hours actually go.
data-room files byte-verified against their claimed types
confirmed mislabeled files — and unreadable files — surfaced by name, not skipped
red flags, every one carrying a citation back to the seller's own documents
marketed PV10 labeled unverifiable — the native reserve databases it rests on were never delivered
The Standard, Rule 1 — applied to the seller's headline valuation.
What we verify
Deterministic where determinism matters, cited everywhere, and honest about what cannot be verified. This is verification infrastructure, not another AI reader.
Every file byte-verified against its claimed extension and format. Mislabeled, corrupted, and unreadable files reported by name and count — before your team bills an hour against them.
Figures and summaries that depend on underlying data — native models, ledgers, executed versions — are checked for their sources. No source: the claim is labeled unverifiable. Never estimated.
Every finding carries its citation — file, location, context. A finding the engine cannot trace is a finding it does not make. Your memo inherits the receipts.
Late additions, replacements, and quiet removals are flagged as they land — the supplemental 10% is exactly where signals get buried. The room you verified is the room you're still looking at.
Findings are computed and receipted first; the narrative renders only from receipted findings. No assertion without a backing receipt — a violation is a build error, not a style issue.
Search and ranking surface candidates across the room at review speed. Verification is always deterministic — retrieval decides what you read first, never what the answer is.
Engage
Legacy data rooms charge by the page and exceed their own quotes 2–10× (SRS Acquiom analysis of 3,800+ deals). Verification shouldn't work that way.
Buyer-side verification of one data room, flat fee, published price.
For M&A practices and diligence teams running continuous deal flow.
The room verified before bidders see it — because a clean room closes faster.
Send us the room. We return the integrity report, the red-flag register, and the citations — every finding traced, every gap labeled, before you sign.
Currently onboarding early engagements. Energy, legal, and middle-market M&A first.