Point Bassethound at any domain and it brings back the whole company: the tech they run, who they are, and whether they are actually building with AI. Five layers, one call, no setup.
claude mcp add --transport http bassethound https://mcp.bassethound.ai/mcp Connect the hosted endpoint in Claude or ChatGPT, or drop one line into Cursor, Windsurf, or your terminal. Nothing to host, no key to babysit.
$ claude mcp add --transport http bassethound https://mcp.bassethound.ai/mcp$ codex mcp add bassethound --url https://mcp.bassethound.ai/mcp{
"mcpServers": {
"bassethound": {
"url": "https://mcp.bassethound.ai/mcp"
}
}
}{
"mcpServers": {
"bassethound": {
"serverUrl": "https://mcp.bassethound.ai/mcp"
}
}
}$ code --add-mcp '{"name":"bassethound","type":"http","url":"https://mcp.bassethound.ai/mcp"}'https://mcp.bassethound.ai/mcpIn Claude, open Settings, Connectors, Add custom connector, and paste the URL. Setup guide
https://mcp.bassethound.ai/mcpIn ChatGPT, open Settings, Connectors, and add the URL as a custom connector. Setup guide
{
"mcpServers": {
"bassethound": {
"url": "https://mcp.bassethound.ai/mcp"
}
}
}
Then just say: sniff stripe.com. Digby takes it from there.
Most tools just tell you what a website is built with. Bassethound checks five layers and connects them into a picture of the whole company.
Frameworks, analytics, CMS, payments, and CDN vendors. The tools a company actually ships with.
DNS, hosting, mail providers, and certificates. Where the company runs and how it is wired.
Size, sector, location, and funding. The shape of the business behind the domain.
The AI stack it actually runs: model providers, vector stores, MCP endpoints, and llms.txt. Shipping, experimenting, or none.
Exposed surface, response headers, and hygiene signals worth a second look.
The playful language stops at the workflow. The deliverable is a serious, source-linked dossier your team can act on and forward without editing.
Every claim traces back along its trail of evidence. Nothing asserted without a citation.
Readable prose a person can skim, and structured JSON an agent can parse. Same sniff.
Titled Company Intelligence Dossier, clean enough to send to a customer, exec, or investor.
Set a watch on a company and Bassethound keeps sniffing. When something meaningful changes, that is a signal, and your agent is the first to know.
The AI a company actually runs, every finding tied to its source, from one hosted endpoint your agents call mid-task. Four ways Bassethound is built differently from an enrichment API.
Bassethound fingerprints the AI stack behind a domain: model providers, vector stores, MCP endpoints, and llms.txt, then calls it shipping, experimenting, or none. The one signal every other enrichment tool skips, correlated with four more layers into a single read.
Digby is honest about uncertainty. Every finding carries its evidence, and a maybe is labeled a maybe. No confident nonsense.
A tool your agents call mid-workflow, not a dashboard you log into.
One hosted endpoint at mcp.bassethound.ai. Nothing to deploy, no keys
to babysit.
Free to start and keyless. When you outgrow the free tier you upgrade from inside your agent, no seats and no dashboard to manage.
$0 forever
For a few lookups a day.
$49.99 per month
For builders running frequent analyses.
$149 per month
For teams and production workloads.
Custom contact us
For committed volume and integrations.
Paid plans are billed by Stripe and turned on from inside your agent: connect
Bassethound, run the upgrade tool, and your higher limits apply on the
next call. Go over your monthly quota and you drop back to the free 3 a day, never to
zero. Cancel anytime.
One sniff, and the whole company comes back. Sourced, and ready to forward.
