Sixteen Sections
Wall Street’s New First Analyst Is a Chatbot
The IPO window is open again — and a new analyst beat Wall Street to the desk. Before a banker builds the book or a reporter files the story, the buyer types the company name into a chatbox and reads whatever comes back. That answer is the first impression now. This study measures who controls it across 25 recent and pending U.S. IPO candidates.
The pattern is a split screen. AI-infrastructure and crypto names are explained with confidence and tend to be sourced through the company’s own material. Consumer-fintech and design-software names are recognized instantly but explained through the media narrative — often the negative one. A cluster of pending filers are functionally invisible or confused with competitors, precisely when their story is least settled and most consequential.
You can ace the audit, build the book, ring the bell — and still flunk the first question a buyer types into a chatbox.
More Buyers Ask Bots Than Bankers
The behavior shift is now measured, not assumed. In a March 2026 survey of B2B software buyers, G2 found that 51% now begin research with an AI chatbot more often than with Google — up from 29% a year earlier — and that one in three bought from a vendor they had never heard of before the chatbot named it. That is the investor-adjacent audience: the analysts, reporters, recruits, and researchers who form the first read on a company.
Every IPO candidate has spent 12 to 24 months on audits, governance, and an S-1. Almost none have spent a day on how the engines will explain them. The S-1 is written for the SEC. The chatbox answer is written by whoever fed the model. When those disagree, the machine tends to follow the press.
How the Index Is Built
This is a directional modeled index, not a logged audit. Each company was modeled across five engines — ChatGPT, Claude, Gemini, Perplexity, Google AI Overviews — against a fixed bank of 60+ investor-intent prompts spanning four families: recognition, accuracy, sourcing, and answer quality. Four dimensions were scored 0–100 and combined into a single AI Readiness Score:
- Recognition — does AI identify the company correctly and distinctly?
- Accuracy — does it explain the business model and financials correctly?
- Source Control — does it lean on the company’s own pages and filings, or on third-party media?
- Answer Quality — is the tone positive, neutral, or confused?
Source Control is the proprietary metric. It separates a company that is merely visible from one that is visible in its own words — the difference that matters most at the moment of listing.
The Leaderboard
The full ranking. Sort any column; filter by status or sector. Every score is a directional modeled estimate.
StatusAllRecently publicPending / expectedSectorResetExport CSV| # | Company | Sector | Status | Recog. | Accur. | Source Control | Quality | AI Readiness | Modeled Share |
|---|
Source Control: OWN = own pages/filings lead · MIXED = balanced · MEDIA = third-party media leads. Quality: POS / NEU / CONFUSED. A clean printable table appears in the Appendix (Section 15).
Infrastructure Owns the Answer
The top of the index is AI-infrastructure and crypto: CoreWeave, Circle, Anthropic, Databricks, Anduril. The engines identify them instantly, explain them correctly, and frequently cite their own technical material and filings. They earned that not through PR but through volume — they published densely about themselves while the market wrote densely about them.
Out-write the market, and the machine repeats you. Get out-written, and it repeats them.
The lesson generalizes: density of authoritative, primary-source content is the strongest predictor of a controlled answer. The leaders earned it by circumstance. Every other filer has to build it deliberately.
Famous — and Defenseless
The middle tells the more useful story. Klarna and Figma are recognized perfectly and explained almost entirely through the post-IPO decline narrative the press built — the drawdown, the lawsuits, the “valuation hangover.” None of it is factually wrong. All of it is uncontrolled.
This is the most expensive failure mode in the study: high recognition, low source control. The company is unmistakably visible and the narrative is already set, with nothing pointing back to its own framing. A newly public company in this position is being understood on someone else’s words.
The Billion-Dollar Ghosts
At the bottom sit pending filers that are invisible or actively confused — blended with better-known competitors, handed stale funding figures, or returned with hedged “I’m not certain” answers. Entrata, Crusoe Energy, Genesys, and Lime land here: real businesses with real filings whose AI footprint hasn’t caught up to their ambitions.
It is the cheapest problem to fix and the most dangerous to ignore. A confused answer at the moment of filing is a confused first impression for every reporter and investor who checks — and it compounds, because the engines cite one another.
The index in miniature — one company at each end of the Source Control spectrum, and one in the dangerous middle.
