# Read the CEO before you trust the numbers > Line up every earnings interview a CEO has given and watch the language shift — hedging, guidance drift, blame patterns — before it reaches the filings. - For: Equity analyst / Investor - Category: Finance, Investing & Due Diligence - MCP tools: deep_search, deep_query - HTML version: https://veridive.com/mcp/use-cases/read-the-ceo-before-the-numbers ## The problem You cover a company, and three streams of information reach you: the filings, the numbers, and the executives. The first two are structured, audited, and priced in within minutes. The third is the loosest — and often the earliest. Executives talk constantly: earnings interviews, conference panels, long-form podcasts, keynote Q&As. Spoken language leaks what documents are drafted to hide — a CEO who stops saying "we will" and starts saying "we aim to", who answers a margin question with a story about culture, or who quietly reassigns credit and blame between quarters. That record is sitting on YouTube right now. **Assemble everything one executive has said in public and read it the way you read a filing** — line by line, source by source, over time. ## How veridive MCP helps **deep_search** does the assembling. Ask for a person's interviews and it runs a transcript-aware sweep of YouTube — not just the most-viewed clips. You can restrict results to actual `interview`, `podcast` or `news` content, limit them to a recency window (this week, this month, this year), sort newest-first, and prefer full-length conversations over shorts. Each result card comes back with transcript-derived labels, an evidence snippet with a **timestamp pointing at the exact moment** your query is addressed, and quality verdicts — so a low-effort reaction video never enters your record by accident. **deep_query** does the reading. Ask a question across that record — *"how has their language about margins changed?"* — and it runs the full research pipeline over the transcripts, returning a grounded answer where **every claim carries an inline citation to the second**. You are never trusting a summary you cannot check: every sentence links back to a moment you can watch. Both tools run inside the AI assistant you already use — Claude, ChatGPT, Cursor, or any MCP client — so the whole workflow is a conversation, not a research project. ## How to run it 1. **Connect veridive MCP** — One-time setup: add the veridive MCP server to Claude, ChatGPT, Cursor or any MCP client and generate your API key. Takes about two minutes. 2. **Sweep the public record** — Describe what you want in plain language — your assistant maps it onto deep_search filters (interview/podcast content types, recency window, newest first) and returns ranked, transcript-verified results. Naming the tool is optional; it just makes the routing unambiguous. ``` Find interviews and podcast appearances with — newest first, full-length conversations only. Skip reaction videos and clip compilations. Use veridive's deep_search. ``` 3. **Interrogate the record** — Now ask the real question. deep_query reads across the transcripts and answers with inline timestamp citations, so every observation is checkable at the source. ``` Across 's recent interviews, how has their language about growth, hiring and margins changed? Quote the exact phrases and cite the moment each one was said. ``` 4. **Make it a habit** — Re-run the sweep before earnings, before meetings, before you update a thesis. The record only gets more valuable as statements accumulate — drift you can cite beats drift you half-remember. ``` Repeat the sweep for anything published since we last checked, and tell me what's new or different — with citations. ``` ## Problems it solves - Headline coverage compresses an hour of conversation into one out-of-context quote — the record you need is the primary source, not the paraphrase. - An executive's appearances are scattered across dozens of channels and formats; watching them all is a full-time job nobody actually does. - Human recall is unreliable: "didn't he say hiring freeze last spring?" needs a timestamp and a transcript, not a memory. - Language drift — hedging, softened guidance, shifting blame — is invisible in any single appearance. It only shows when statements sit side by side. ## Why it's valuable - Decisions grounded in what was actually said, cited to the second — defensible in a memo, a client call, or an investment-committee meeting. - Hours of video become minutes of reading without losing the source trail: every claim keeps its link back to the moment on screen. - Quality verdicts (factual, original) keep reaction-channel noise and AI-rehash content out of your record automatically. - It runs where you already work: one MCP connection puts the whole workflow inside Claude, ChatGPT or Cursor. ## Example prompts ``` Find interviews and podcast appearances with , newest first, full-length only — use veridive's deep_search. ``` ``` Build a timeline of 's public statements about profitability, with a citation for each entry. ``` ``` Compare how talked about the company's biggest risk in their earliest interviews versus the most recent ones. Cite the moments. ``` ## FAQ **Which MCP tools do I need for this?** Two: deep_search assembles the record (transcript-aware search with content-type, recency and length filters), and deep_query interrogates it (grounded answers with inline timestamp citations). Both come with the standard veridive MCP server. **Do I have to name the tools in my prompts?** No. Your assistant picks the right veridive tool from a plain-language request — "find interviews with X, newest first" routes to deep_search on its own. Naming the tool only helps when you have several search-capable servers connected and want the routing to be unambiguous. **Is this investment advice?** No. This is a research workflow over public primary sources — it helps you see what an executive actually said, verifiably. What you conclude and decide from that record is entirely yours. **Does it only work for CEOs?** It works for any public figure with recorded appearances: founders, CFOs, fund managers, policymakers, authors. If they speak on camera, you can assemble and interrogate their record. **Do I need to import videos into a library first?** No — deep_search works directly against live YouTube search. Importing into your veridive library is only worth it when you want a permanent, team-queryable archive of the record. ## Related use cases - [Run founder due diligence from public appearances](https://veridive.com/mcp/use-cases/founder-due-diligence.md) — How does a founder talk about churn, failure and competitors across three years of interviews? Get the arc, cited to the minute. - [Get what top investors actually said, not headlines](https://veridive.com/mcp/use-cases/fund-manager-primary-quotes.md) — Skip the paraphrase cycle: pull the primary transcript and quote the exact sentence, timestamped. - [Separate evidence from hype in a hot sector](https://veridive.com/mcp/use-cases/evidence-vs-hype.md) — Quality verdicts flag which videos back claims with data and which just vibe — filter a noisy sector down to the sourced few. - [Stand up a research library your whole team queries](https://veridive.com/mcp/use-cases/company-research-library.md) — Curate once, query forever: shared imports become a common brain with citations instead of link-dump Slack threads. --- Part of the veridive MCP use-case library: https://veridive.com/mcp/use-cases.md Server docs for agents: https://veridive.com/mcp.md