For investors and for founders raising

Know the company.

An analyst’s work, without needing to hire an analyst.

Upload a deck. Verdict tells you whether the company is worth a meeting, and why. The one shown is real.

Run your deck

About ten minutes.

Only your account can open it. Your deck never trains an AI.

Read the published memos

Verdict · memoFTX · Growth

FTX

PASSConfidence · High

Do next

Request audited financials and custody proof.

Pass: custody opacity is the dealbreaker.

The deck discloses no custody mechanics and no round terms. At $14.7B average daily volume, the absence of any statement on whether customer assets are segregated from FTX operating capital is the central risk, not a diligence footnote, and no investor can responsibly commit capital without audited proof of segregation and a term sheet.

The one blocker

Customer fund segregation never disclosed in deck

StageGrowth

SectorFintech / Cryptocurrency Exchange

Average Daily Volume$14.7B

Annualized Revenue Run Rate$1.2B

Flags

3 surfaced · 1 critical

Slide 3
Slide 3 states $800M estimated profit in 2021 with asterisk noting 'numbers are approximate, based on recent performance. They do not incorporate any additional growth.' This qualifier undermines the precision of the headline traction figures.

General verdict

Biggest risk

The deck never discloses whether customer assets are segregated from FTX operating capital, and at $14.7B average daily volume that omission makes the solvency question unanswerable from the deck alone.

Best reason

FTX's founding team built the trading infrastructure at Alameda Research before building the exchange, giving them a founder-operator fit that is rare in crypto and that produced $1.2B in annualized revenue by 2021.

Would change mind

Audited financials confirming the $800M profit figure, a written proof of customer fund segregation from a Big Four auditor, and a term sheet with round size and governance terms would reopen this as an Invest conversation.

Why this verdict

01
Custody and flow-of-funds mechanics entirely absent from the deck, the single most dangerous undisclosed fact at this volume scale
02
Profit figures explicitly approximate and unaudited, with the deck's own disclaimer undermining the headline
03
No round terms disclosed, round name, amount, valuation, and governance rights are all absent

Read the whole memo

Published memos

Same four axes. Different numbers.

Passing on FTX is easy in hindsight. The harder test is whether the same scale still hands an A to a company it is turning down. Both memos are public and whole, so you can check the call yourself.

Grades on four axes for the two memos below, FTX and Theranos, with each memo’s verdict and its next step
Company	Unicorn opportunity	Capital efficiency	Team velocity	Moat durability	Memo
FTX	B+7/10	A8/10	A9/10	B5/10	Whole memo
PASS3 flags · 1 criticalDo next Request audited financials and custody proof.
Theranos	B5/10	C+4/10	B5/10	C3/10	Whole memo
PASS7 flags · 2 criticalDo next Demand independent analytical validation before any capital.
The scale	F1-2C3C+4B5-6B+7A8-9A+10

Written in public

“Everyone gives you a reaction. Nobody gives you a ruling.”

Kevin CarterCoach's Corner

Ethan TylerFounder, sent.bio
“[It] was brutally honest and answered EVERY question I had. If I had lunch with 1,000 Joe Shmoe VC's it still wouldn't have been as valuable.”
Akwarandu PrincewellFounder, SCopyTrade
“I tightened everything based on that feedback. Started sending the revised version out. People started responding. Not polite passes. Real conversations.”
Elijah MayFunded HouseAdvises Verdict
“We've already integrated Verdict into Venture Scoreboard by Funded House® and are using it to help startups and investors both get clarity on investability.”

Why not a chat window

What a chat window cannot do.

A chat starts from nothing every time.

It shows its work
Every number links back to the page it came from.
It learns
It checks its own work after every memo and remembers what it got wrong.
It knows your fund
After a few memos on a fund, it reads new decks the way that fund would.

Frequently asked

The questions everyone asks first.

How is this different from ChatGPT, Claude, or Gemini directly?

A climate-tech deck claimed 450 terawatts of waste heat on slide 3, and nobody who built it caught the error. Terawatts measure power, not energy, so the figure was impossible. Verdict caught it on the first run. A general assistant returns the answer a reader will accept. Verdict looks for the thing that kills the deal.

What about hallucinations?

Every figure links back to its source, the exact deck page or web page it came from, so any claim can be checked in one click. Verdict also writes two verdicts. One on the company, one on its fit with a particular fund's thesis, so it has to commit to each.

How does sharing a memo work?

Every memo has a read-only link, and whoever you send it to needs no account. Paid plans export a PDF with a confidential mark and every source listed at the back.

What does a founder get from running their own deck?

The memo a partner would write about you, before you walk into the room. It sizes your TAM against real sources and names the competitors left off your slide. Coaching mode turns each finding into a fix you can make before the meeting.

Your data is safe

Every deck you upload is sensitive. Here is exactly what happens to it.

Private to you

Files belong to the account that uploaded them, and Verdict looks them up by that account every time it reads them.

Never trained on

That covers the memo and any feedback you write, not only the deck. Anthropic runs the model behind Verdict, and its commercial terms forbid training on anything a customer sends.

Encrypted

Encrypted at rest and in transit. The exact standards are on the security page.

Yours to delete

Deleting a memo removes the deck file in the same step. Anything you keep is deleted after 90 days, or 24 hours from upload if you marked the deck sensitive, whether or not the memo finished.

Read the full security posture, with retention tiers and sub-processors.

Run a deck and see what comes back.