Point your AI at a lead — get back a fit score and a ready-to-send email.

Not a mockup — a recorded run: an agent called these tools over real data, and this is what came back. It scores the person and drafts the outreach, then profiles the account behind them. Same data layer, two doors: MCP for your AI, REST for your code. On the right, the artifact it builds as it goes.

01

Score the lead, draft the outreach

analyze_contact — a fit score, a written email + LinkedIn, and the do/don't. email_intel checks the address first.

score a lead
← the run — every call and the line the agent readsthe scorecard it builds — score, drafts, the play →
claude — aidenix-mcpconnected
agent's artifact
🎯
the scorecard lands here
the agent is checking the ICP and the address first — those answer in a line, so they stay on the left
02

Profile the account behind them

company_signals, then the same call one level deeper — is now the time, with what, and through whom.

ask about
← the run — every call and the line the agent readsthe account panel it builds — card, momentum, attention →
claude — aidenix-mcpconnected
agent's artifact
⚡
the agent hasn't called a tool yet
watch it think and reach for Aidenix on the left

Every endpoint, one token

analyze_contactICP fit score + a written email & LinkedIn, with the play
email_intelis the address live, and is the person still there
email_intel_bulkthe same question across a whole list, in one call
person_signalswho they are, what they publish, when to reach them
company_signalsaccount timing & momentum — and, on request, who the team watches
analyze_contacts · batchesscore whole lists, then export the ICP-60+ ones
build_icp_profilepoint it at your own site — it becomes what every score weighs against
list_icp_profileswho you sell for — switch it and every score reweighs
get_account_contextyour plan and what is left of the quota
add_opt_outsomeone asks to be left alone — record it from the chat

How to connect — this is a real MCP server

{
  "mcpServers": {
    "aidenix": {
      "url": "https://mcp.aidenix.com/mcp",
      "headers": { "Authorization": "Bearer ‹token›" }
    }
  }
}
Get a token and connect →
That URL is live — it is the server this demo was recorded against.

What the agent sees

Every tool returns two things: a human-readable takeaway (the agent reads it and reasons on it) and the full structured block (for its workflow). No target-company employee names in the response — only business signals.

10.8B people & company docs · scoring and drafts weighted by your saved ICP · LinkedIn snapshot 2026-06, sources coresignal + contora + discolike + progai.