Let your agent read the heat and hand it back.
HeatCheck exposes an MCP server so your AI assistant can pull the scored shortlist, explain a score, and draft the tasks, over a secure connection.
The problem
Reps increasingly work through an assistant. If the heat data is not reachable from there, it is one more tab they will not open.
What it does
Ask in plain language
“Who crossed into SQL this week?” returns the scored, itemized answer.
Scoped access
The agent acts as the signed-in member, with that rep’s permissions.
Draft, not send
The agent can draft tasks; sending stays a human action.
How it works
Connect
Add the MCP server in your assistant as a connector or with an API key.
Query
The agent reads scores, tiers and audit trails.
Hand back
It returns a shortlist and drafts, with a link into HubSpot.
Write actions are gated behind a separate permission, so read-only agents can never send on their own.
The heat should be one question away, in whatever tool your reps already use.
The short version.
How does an agent connect?
Over MCP, either as an OAuth connector in Claude or ChatGPT, or with an API key for coding agents.
Can the agent send emails?
No. It can read scores and draft tasks; sending is gated behind a separate write permission and a human.
Whose data can it see?
It acts as the signed-in member, scoped to that rep’s HubSpot permissions.
See who’s hot right now.
Early access is open for a small design-partner group.