The anatomy of a Verdict report, shown on an illustrative subject: verdict on page one, red flags ranked, every material claim sourced or labelled. For a real, unedited report built entirely from public sources, see the published sample.
One company per report · ten fixed sections · every claim cited or marked as inference
Verdict on page oneEvery claim checked at sourceTen sections · signed
Annotated walkthrough
The report, with the parts pointed out.
This is page one, on an illustrative subject. The verdict sits up front, the reasons stack beneath it, red flags are ranked, every claim sourced. Five parts, pointed out.
Illustrative structure on an invented subject. Halden Robotics is invented to explain how a report is laid out. Any resemblance to a real company is coincidental, and findings here are illustrative, not statements of fact about any company. The published sample is a real, unedited report from public sources.
Due diligence report · page 1VRD-2026-0625 · illustrative
Halden Robotics, Inc.
Prepared by Verdict, a Lucid Labs product · Report date June 25, 2026 · Stage Late‑stage private (Series C) · Decision context Invest / secondary purchase · Funding raised ~$240M across 6 rounds
1 · Executive summary + verdict
1
VERDICT, PROCEED WITH CONDITIONS
12Growth is real and accelerating. ARR compounded from ~$6M to ~$55M in 24 months, with a verified 50% single-month jump on the pricing shift.
2The IP overhang is existential, not cosmetic. An unresolved IP-assignment dispute with a former founder, layered on heavy top-customer concentration.
3Valuation premium demands unproven execution. ~$2.1B post against ~$55M ARR is roughly a 38x multiple, versus 10 to 20x at SaaS IPO.
38 · Risk register, ranked by severity
CriticalCustomer concentration plus an unresolved IP-assignment dispute
A founder-era IP-assignment dispute, unresolved, could cloud title to core patents and amplify top-customer concentration risk. [2][3][4]
Flag 1
4Conditions for proceeding: confirm IP-assignment release and patent-title clearance in the data room; validate whether the top three customers sit above or below 40% of ARR; stress-test the multiple at 25x ARR for a downside case. [1][5]
5··· Sections 2 to 10 continue. Around 3,300 words, each claim carrying a citation or an inference label.
Five parts, pointed out
1
Verdict on page one
Proceed, conditions, or pass, stated before the evidence, not buried behind it.
2
The three reasons behind it
The call justified in three lines you can scan before you read a single section.
3
Red flags, severity-ranked
The risk register leads with what matters most. It's the section read first.
4
Conditions to proceed
The exact things to confirm before you act, written as a checklist.
5
Every claim sourced or labelled
Material facts carry a numbered citation; thin signals are flagged, not guessed.
Custom scope
Need a dimension we don't list? Ask for a custom section, any angle, scoped to your deal, in the same sourced format.
~3,300 words · 10 sections · every claim cited or marked as inference. Illustrative sample, fictional company: Halden Robotics is invented to demonstrate report structure. Findings are illustrative opinion, not statements of fact, and not affiliated with or endorsed by any company.
Same ten dimensions on every company, so two reports read side by side. Tap through a few to see the actual shape of each, on the illustrative subject.
01 · Executive summary + verdictPage 1
VERDICT, PROCEED WITH CONDITIONS
1Growth is real and accelerating. ARR compounded ~$6M to ~$55M in 24 months, a verified 50% single-month jump on the pricing shift.
2The IP overhang is existential. An unresolved IP-assignment dispute with a former founder, layered on top-customer concentration.
3The valuation premium demands unproven execution. ~$2.1B post on ~$55M ARR, roughly 38x versus 10 to 20x at SaaS IPO.
08 · Risk registerRanked by severity
CriticalCustomer concentration plus an unresolved IP-assignment dispute
A founder-era IP-assignment dispute could cloud title to core patents and amplify top-customer concentration risk. [2][3][4]
Flag 1
HighRevenue quality unconfirmed below the ARR headline
Single-month step-up not yet reconciled to net revenue retention or gross margin. [5]
Flag 2
09 · Questions for managementTake into the call
Q1Can you show the IP-assignment release and patent-title chain of custody in the data room?
Q2What share of ARR do the top three customers carry, and how has that moved over eight quarters?
Q3Does the 50% single-month jump hold on net revenue, or is it gross bookings before churn?
10 · Sources & confidenceEvery claim traceable
[1]SEC EDGAR: Form D filings, funding rounds and post-money markers. Confirmed.
[2]Court docket: founder IP-assignment dispute, filing and status. Confirmed.
[3]Customer concentration, triangulated from case studies and hiring signals. Inference, labelled.
The alternative
A verdict, or a guess.
What you'd otherwise do with the same decision: pay a firm for billed hours, or trust your gut. Here's the same call, three ways.
Attach a deep-dive to any report: AI & tech-stack dependency, customer concentration & churn, or key-person risk.
Run blind on a deal that later went bankrupt, this engine said do not close. The buyer closed. Every issue public records could reach, the blind lanes surfaced, and the full finding-by-finding scoring is published. Read the backtest.
After the report
The report is the start, not the end.
A Verdict is a snapshot of the day it's signed, and businesses move. Watch keeps the file live: every month I re-run the same diligence, compare it against your report, and send a short memo on exactly what changed.
New filings, new risks, fresh red or green flags. You stay current without re-buying the whole investigation.
A monthly memo
A short "what changed" note, every month, in plain English. No dashboard to check.
Deltas on your claim table
Each change measured against the claims in your original report, sourced the same way.
A heads-up when it's material
The moment something moves that actually affects the call, you hear it, not next quarter.
Watch
Keep watching after delivery.
The light monthly version of diligence, ongoing. Distinct from the one-off 90-day refresh, which is a deeper re-run of the whole pipeline. Offered when your report lands, so you decide once you've seen the file.
Built from primary sources. Honest about the gaps.
Every verdict is assembled from the public record, then graded by how solid each claim is. Confirmed facts carry a citation; inferences are labelled; thin signals are flagged, never dressed up as fact.
Primary
Filings, registries, official statements. A company filing, a state or federal registry record, a named person on the record.
Reputable secondary
Established press and trade reporting. Used to corroborate, dated and named in the citation.
Market research
Third-party research firms. Market size and comparable data, cited to the firm and the year.
Secondary
Analyst and aggregator estimates. Carried as estimates, never upgraded to fact.
Inference
Our judgment from sourced facts. Labelled as inference in the report, with the reasoning shown.
Unaudited figures carry a symmetric skepticism band: the seller's numbers and the comparables get the same discount. Claims made by interested parties are flagged as such.
Confirmed, sourced and citedInference, derived and labelledUnverified, thin and said so
Exhibit · from the published sample
The missing-data register: what a real buyer must still obtain before closing.
Every report ends with one. This is the register from the Anysphere (Cursor) sample: ten things the public record could not verify, named instead of papered over.
Gross margin by tier and usage cohort. The single most important operating unknown.
Net revenue retention, split by cohort. Individual vs. enterprise; currently undisclosed, and we infer divergence.
Model-vendor contracts. Pricing protection, access guarantees, exclusivity, and term, for Anthropic and OpenAI.
Proprietary-model substitution rate. Composer's share of inference tokens and its cost-per-token delta; drives both margin and the disintermediation trigger.
Cap table and liquidation-preference waterfall. Participation, multiples, seniority by round, and the implied common-holder step-up under the all-stock structure.
Cohort retention curves and CAC payback. By channel, individual vs. enterprise.
Top-N customer concentration. Top-10-logo share of ARR, and the share held by investor-customers.
IP and provenance file. Base-model license compliance, export-control posture, and the reality of any related-party training-compute arrangement.
Each report closes with a Sources & Confidence section: the full citation list plus analyst notes on exactly where the gaps are. You see how solid the ground is before you act on it.
The full desk
Verdict is the anchor of a four-seat desk.
The same sourced, signed discipline runs the length of the deal, from the first listing to the final price.
One company per report. Verdict up front, ranked red flags, every material claim sourced or labelled. Delivered as PDF, editable Word, a 1-page brief, and an IC-ready memo, no retainer and no call required.
Ninety days on, a refresh re-runs the pipeline and lands as a diff against your original.