MCP Explained in 5 Minutes: A Guide for HubSpot Leaders Who Don't Code
This is the 5-minute version. If you’re a VP of RevOps, an agency owner, or a HubSpot Solutions Partner being asked “are we doing anything with AI?” - forward this, read it on the train, then read the full executive guide when you want the details.
What is MCP - in one sentence
MCP (Model Context Protocol) is the open standard that lets AI assistants like Claude and ChatGPT talk to tools like HubSpot. Anthropic released it in late 2024. Every major AI tool has since adopted it. It’s the USB-C of AI tooling.
HubSpot now ships its own first-party MCP - the Remote HubSpot MCP Server, generally available since 2026-04-13 at https://mcp.hubspot.com. It’s free, works across all hubs and tiers, and lets any AI client your team already uses (Claude Desktop, Cursor, VS Code, ChatGPT) read and write against the standard CRM via OAuth 2.1. That’s the first-party option most RevOps teams should try first. See the HubSpot MCP pillar for the full technical walkthrough.
That’s the mechanism. You don’t need to understand how MCP works. You need to understand what it unlocks.
Why it matters to your P&L
Most of your ops labour cost is in repeatable work:
- Auditing a portal before onboarding a client: 2 hours, done by a senior.
- Building a segmentation list from a plain-English request: 30 minutes, done by the admin.
- Standing up a renewal system: a full afternoon.
- Weekly rep record-hygiene reviews: a couple of hours per rep per month.
None of these are hard. They’re slow because the operator is the bottleneck between HubSpot’s UI and what they’re trying to do with the data. MCP-based tooling collapses that bottleneck. Not by replacing the operator - by letting them skip the clicking.
In practice: portal audits move from 2 hours to ~10 minutes. A Human-Approved Write Plan can draft 200 clicks of setup in one reviewable step. Adding 9 segmentation lists goes from an afternoon to 2 minutes.
The ROI is cycle-time reduction, not headcount reduction. Same team, 5 - 10× the output.
The operating model: Humans AND AI
The framing that survives the board meeting is humans AND AI - not autonomous agents. Every change the AI recommends is reviewed and approved by a person before it touches production data.
Concretely: the AI drafts a Write Plan - a reviewable sequence of changes - and the operator reads, edits in conversation, and confirms in a single step. The plan document itself becomes the audit record. No change reaches the live portal without a human signing off.
It’s the difference between “AI in the loop” and “AI running the loop.” You want the first.
Cost framing
- Free - daeda-mcp npm package + any AI you already have. Good for individual operator evaluation.
- $125/month - Daeda AI Consultant Plan: 1 user, 5 portals, Write Plans, Workflow Actions, Autofill. Breaks even the first time it saves a 2-hour audit.
- $375/month - Daeda AI Agency Plan: unlimited users, 5 portals. Built for teams running the same workflows across the estate.
- Model costs - bring-your-own-key via OpenRouter, paid directly at cost. No markup.
Per-seat pricing is the wrong frame. This is a capability layer on top of tools your team already uses.
What to do this week
Three questions for your Monday meeting:
- Which job is our team tired of re-doing manually across portals? That’s the first workflow to ask the AI to draft.
- Who on the team is already using Claude, ChatGPT, or Cursor every day? They’re your proof-of-value operator.
- What does a 30-day win look like? Pick one audit, one build, one recurring task. Measure before/after. Don’t start with a deck.
When you’re ready to scope it, the full executive guide has the longer walkthrough, and the HubSpot MCP pillar page is the technical companion.
