Named verifier · Accountability
The name on every verdict.
Every PRISM report and Sentinel verification ThriveFinity issues carries a named human signature. This page tells you who that person is, what they check, and what they are accountable for.
The process
The verification process — six steps
Every PRISM Pro and Ultra report, every Sentinel verification, follows this protocol before Pranav signs.
Brief intake & claim extraction
The submitted brief, deck, or document is parsed to extract every assertion that could be challenged: market size figures, competitive claims, benchmarks, pricing rationale, regulatory assumptions.
PRISM lens assignment
Each extracted claim is assigned to one or more of the 12 PRISM lenses: market sizing, competitive moat, unit economics, demand validation, regulatory risk, distribution, team execution fit, and five more.
RAG retrieval from licensed corpora
Claims are cross-referenced against licensed, datestamped sources via retrieval-augmented generation. Web-scraped results are not used. Every retrieved citation is checked for publication date, methodology, and primary sourcing.
Adversarial stress-testing
Pranav reads every AI-generated draft and tries to break each claim before the report does. For Pro and Ultra tiers, 15–60 stress scenarios are run against the business model assumptions. If a claim survives stress-testing, it goes to verdict.
Verdict assignment & sign-off
A PRISM verdict (GO / CONDITIONAL GO / PIVOT / DEFER / KILL) is assigned based on aggregate lens scores. The verdict is signed by Pranav with a date stamp. For Sentinel reports, the specific verified and disputed claims are itemised with source trails.
Delivery & 30-day guarantee
The signed report is delivered within 24–48 hours depending on tier. The 30-day unconditional refund guarantee activates on delivery. If a cited claim is later found to be wrong, the errata process opens automatically.
Why AI alone is not enough
AI accelerates. A human is accountable.
AI retrieval and synthesis have made it possible to cross-reference hundreds of licensed sources in seconds. That speed is real, and we use it. But AI tools cannot be held professionally accountable for a wrong verdict. They do not have a name on the document. They cannot be contacted when an investor challenges a claim. They cannot issue a refund.
The Sentinel verification process — which underpins PRISM and pitch deck verification — uses AI at every retrieval and synthesis step. What it does not use AI for is the final verdict sign-off, the adversarial stress-testing, or the source verification gate. Those steps require a human with skin in the game.
Errata & corrections
If a verdict or report Pranav has signed contains a factual error — a wrong citation, a miscalculated score, a misidentified competitor — we correct it publicly under the errata policy. The correction is logged, timestamped, and noted on the verdict distribution page. We do not silently edit published reports.
Read the errata policy →