20,000+ executives · 1,000,000+ quotes · full transcripts · 4 MCP tools · 30-second install
Ask any AI agent "what has Jensen Huang said about AI infrastructure?" and you'll get hallucinated quotes, misattributed soundbites, and links to SEO-spam blogs. The web is full of fake CEO quotes — and LLMs eat them all.
CEOInterviews fixes this. Every transcript is AI- and human-verified to the original primary source: the actual interview, podcast, earnings call, or press conference. No fabrications. No misattributions. No "as an AI language model, here's a plausible-sounding quote."
| ❌ Other "executive quote" datasets | ✅ CEOInterviews |
|---|---|
![]() |
![]() |
| Unverified blog repost · No source URL · Misattributed quotes | AI + human verified · Original primary source · Speaker-authenticated |
This MCP server exposes that database to Claude, ChatGPT, Cursor, Windsurf, and any MCP-compatible agent — through 4 simple tools. Connect it once and your agent gains a research analyst with a perfect memory of what every CEO, CFO, central banker, and AI founder has actually said on the record.
claude mcp add --transport http --scope user ceointerviews https://ceointerviews.ai/mcp \
--header "Authorization: Bearer YOUR_API_KEY"That's it. Open Claude Code, type /mcp, and you're connected. Grab a free API key →
Don't use Claude Code? Scroll to Connect any MCP client — Cursor, Windsurf, Claude Desktop, ChatGPT Agents, and raw HTTP all work.
Open your agent and ask any of these out loud:
This is what every "research GPT" promises and almost none deliver. The data layer has to actually exist — and be clean — first.
Here's what a real Claude Code session looks like once you've connected the MCP server. The user asks a single question — Claude discovers the tools, chains two calls, and answers from verified primary sources with citations.
❯ What has Jensen Huang said recently about AI infrastructure
spending? Cite the actual interviews.
⏺ ceointerviews - get_transcripts (MCP)
entity_name: "Jensen Huang", keyword: "infrastructure spending",
filter_after_dt: "2026-01-01"
⎿ 4 transcripts returned
⏺ Across four interviews from Feb–Mar 2026, Huang's thesis is that
AI CapEx isn't a bubble — it's a substitution of existing IT spend
into a more productive form, with the real expansion still ahead.
▸ Morgan Stanley TMT Conference — Mar 18, 2026
"The entire internet industry could take 100% of their CapEx
and make it AI because it's better — we've proven it to be
better."
"Trillions of dollars… that's exactly where all this is
going to come from."
↳ youtube.com/watch?v=xv7UVAfyebk
▸ CNBC Squawk Pod — Feb 26, 2026
"They're supposed to come up with a combined $670 billion
to build out just their AI infrastructure."
↳ youtube.com/watch?v=poS_PlIIx1c
▸ The AI Factory keynote — Feb 6, 2026
"The IT industry is about a trillion dollars… yet the
economy of the world is about a hundred trillion dollars."
Today's IT spend is the floor, not the ceiling.
↳ youtube.com/watch?v=Qt-RFRr5N2I
Every quote above is pulled from a verified primary-source
transcript — the LLM has access to the full transcript,
appearance_date, and rich metadata produced by our team.
No web scraping, no hallucinations.
Without the MCP server, that same question gets you plausible-sounding paraphrases with no sources — or worse, fabricated quotes stitched together from training data. With it, every claim is anchored to a verified transcript.
| 📈 20,000+ executives | Every CEO, CFO, CTO across S&P 500, S&P 1500, NASDAQ 100, and global markets |
| 💬 1,000,000+ quotes | Tagged for notable, controversial, financial-policy sentiment |
| 📜 Full interview transcripts | Earnings calls, podcasts, conferences, press appearances — with source URLs and timestamps |
| 🇺🇸 🇨🇳 🇪🇺 Global coverage | US, China, and Europe-based companies. Public officials and Federal Reserve members. AI startup founders (OpenAI, Anthropic, Mistral, and hundreds more) |
| 🤖 + 👤 AI + human verified | Every transcript is checked for speaker authenticity. Source URLs and publication dates always included |
| ⚡ Streamable HTTP MCP | Hosted endpoint, Bearer token auth, no server to run |
$5 per 1,000 results. $0.005 per result. Billed weekly via Stripe. No monthly minimum, no commitment, no rate-limit gotchas.
| Server URL | https://ceointerviews.ai/mcp |
| Protocol | Streamable HTTP (MCP) |
| Auth | Authorization: Bearer YOUR_API_KEY |
claude mcp add --transport http --scope user ceointerviews https://ceointerviews.ai/mcp \
--header "Authorization: Bearer YOUR_API_KEY"In VSCode, open the integrated terminal (Ctrl+backtick / Cmd+backtick) and run the same command. Then type /mcp in the Claude Code chat to verify.
Paste this into your client's MCP config:
{
"mcpServers": {
"ceointerviews": {
"type": "http",
"url": "https://ceointerviews.ai/mcp",
"headers": {
"Authorization": "Bearer YOUR_API_KEY"
}
}
}
}| Client | Where to paste |
|---|---|
| Claude Desktop | Settings → Developer → Edit Config → claude_desktop_config.json |
| Cursor | Settings → MCP → Add Server |
| Windsurf | Settings → MCP → Add Server |
| Claude Code (manual) | Add to top-level mcpServers in ~/.claude.json |
Point your MCP client at the Streamable HTTP endpoint:
URL: https://ceointerviews.ai/mcp
Auth: Authorization: Bearer YOUR_API_KEY
Returns full verified interview transcripts. Use this first for any question about what an executive has said.
| Parameter | Type | Description |
|---|---|---|
entity_name |
string | Executive name, fuzzy matched (e.g. "Tim Cook") |
entity_id |
int | Exact entity ID from search_executives |
company_name |
string | Company name or ticker, fuzzy matched |
company_id |
int | Exact company ID |
keyword |
string | Search within transcript text (requires entity or company filter) |
filter_after_dt |
string | Only items after this date (ISO 8601) |
filter_before_dt |
string | Only items before this date (ISO 8601) |
page_num |
int | Page number (default 1) |
page_size |
int | Results per page, 1–500 (default 5). Keep small (5–7) — transcripts are large |
Find people. Returns name, title, institution, bio, and company metadata.
| Parameter | Type | Description |
|---|---|---|
keyword |
string | Fuzzy search across name, title, company name, and ticker. Multi-term in any order: "Tim Cook", "AAPL", "CEO Tesla", "devinder kumar amd" |
company_name |
string | Filter by company name or ticker |
gender |
string | M, F, or O |
is_snp500 |
bool | Only S&P 500 company executives |
is_nasdaq |
bool | Only NASDAQ-listed company executives |
is_snp1500 |
bool | Only S&P 1500 company executives |
is_nasdaq100 |
bool | Only NASDAQ 100 company executives |
is_ai_startup |
bool | Only AI startup executives |
is_top_startup |
bool | Only top startup executives |
is_usa_based |
bool | Only US-based company executives |
is_china_based |
bool | Only China-based company executives |
is_europe_based |
bool | Only Europe-based company executives |
is_public_company |
bool | Only publicly traded company executives |
page_num |
int | Page number (default 1) |
page_size |
int | Results per page, 1–500 (default 10) |
Supplementary tool for short notable quotes. Use only when the user explicitly asks for soundbites or exact wording — for general research, prefer get_transcripts.
| Parameter | Type | Description |
|---|---|---|
entity_name |
string | Executive name, fuzzy matched |
entity_id |
int | Exact entity ID |
company_name |
string | Company name or ticker, fuzzy matched |
company_id |
int | Exact company ID |
keyword |
string | Search within quote text |
is_notable |
bool | Only notable/important quotes |
is_controversial |
bool | Only controversial quotes |
is_financial_policy |
bool | Only financial/policy-related quotes |
filter_after_dt |
string | Only quotes after this date (ISO 8601) |
filter_before_dt |
string | Only quotes before this date (ISO 8601) |
page_num |
int | Page number (default 1) |
page_size |
int | Results per page, 1–500 (default 10) |
Look up companies and their classification flags.
| Parameter | Type | Description |
|---|---|---|
keyword |
string | Company name or stock ticker |
is_snp500 |
bool | Only S&P 500 companies |
is_nasdaq |
bool | Only NASDAQ-listed companies |
is_snp1500 |
bool | Only S&P 1500 companies |
is_nasdaq100 |
bool | Only NASDAQ 100 companies |
is_ai_startup |
bool | Only AI startups |
is_top_startup |
bool | Only top startups |
is_usa_based |
bool | Only US-based companies |
is_china_based |
bool | Only China-based companies |
is_europe_based |
bool | Only Europe-based companies |
is_public_company |
bool | Only publicly traded companies |
page_num |
int | Page number (default 1) |
page_size |
int | Results per page, 1–500 (default 10) |
Most users should connect directly to the hosted server at https://ceointerviews.ai/mcp. Self-hosting only buys you the ability to run the MCP transport on your own infra — you'll still need a valid API key to call the upstream CEOInterviews REST API.
pip install ceointerviews-mcp
export CEOINTERVIEWS_API_KEY="your_api_key_here"
ceointerviews-mcpgit clone https://github.com/ceointerviews/ceointerviews-mcp.git
cd ceointerviews-mcp
python -m venv venv && source venv/bin/activate
pip install -e .
cp .env.example .env # then fill in your API key
ceointerviews-mcpdocker build -t ceointerviews-mcp .
docker run -e CEOINTERVIEWS_API_KEY=your_key -p 8001:8001 ceointerviews-mcp| Variable | Default | Description |
|---|---|---|
CEOINTERVIEWS_API_KEY |
(none) | Your CEOInterviews API key |
CEOINTERVIEWS_API_BASE |
https://ceointerviews.ai |
API base URL |
CEOINTERVIEWS_MCP_HOST |
0.0.0.0 |
Server bind address |
CEOINTERVIEWS_MCP_PORT |
8001 |
Server port |
- 🌐 CEOInterviews.ai — main site
- 🔑 Get an API key — MCP landing page
- 📚 REST API docs — full REST reference
- 📄 OpenAPI spec — machine-readable
- 🤖 Model Context Protocol — what's a MCP?
Built for the agents that have to be right.
Get an API key · Star on GitHub · MIT licensed





