Use cases by system8 use cases

What to build on HubSpot with AI, and what Breeze already covers

HubSpot has done more than most CRMs to make itself reachable by AI. Its official remote MCP server signs in with OAuth and reads and writes contacts, companies, deals, tickets, quotes, invoices, orders and engagements, with marketing and content data read-only. Connecting Claude or ChatGPT to HubSpot is therefore a setting, not a project, and I tell people so.

Inside the product, Breeze agents cover customer service, prospecting and data enrichment, with the Breeze Assistant alongside. Agent Hub is a Pro and Enterprise feature, and custom agents consume HubSpot Credits per action. For a team whose answers all live in HubSpot, Breeze is the first thing to try.

The builds in this hub are for work that crosses HubSpot's edge: qualifying inbound leads against capacity, stock or service area held elsewhere, updating the CRM from calls and email after the rep approves, client reporting for agencies that joins HubSpot with the ad platforms, and a scoped MCP server that puts HubSpot and your ERP behind one set of permissions.

What to build around HubSpot

Four things the HubSpot API teaches you early

  • A private app for one portal

    A private app with only the scopes the build needs is the simplest start for a single HubSpot account; a public OAuth app is for software that many portals install. Burst and daily API limits both depend on your subscription tier.

  • Search is the tight spot

    The CRM search endpoint has a far lower rate limit than record reads and caps how many results one query can page through. The agent looks records up by ID or a unique property such as email or domain, and never searches in a loop.

  • Associations carry the meaning

    A deal without its contact and company associations drops out of the reports that matter. Every record the build creates is associated in the same step, using the association labels your team already relies on.

  • Properties are the contract

    Fields such as AI qualification score, AI summary and last AI update make the agent's writes visible and filterable. Property history shows who changed what, so a rep can see and undo anything the agent did.

Breeze or a custom agent: how I decide

  • Breeze when HubSpot holds the answer

    Service answers from the knowledge base and CRM, prospect research and property enrichment are Breeze's home ground. Pay in credits and skip the build.

  • Custom when the answer is in another system

    Stock and pricing in the ERP, subscription status in Stripe, a delivery date in the TMS: Breeze is bounded by HubSpot data, while a custom agent reads those systems through their own APIs.

  • Starter portals and outside channels

    Agent Hub needs Pro or Enterprise. Starter customers, and WhatsApp or phone conversations handled by an agent outside HubSpot, write to the CRM through the API instead.

  • Routing you can audit

    Territory, deal size and language rules run in code, and the model only reads the free text the rules cannot. When a lead lands with the wrong rep, the log shows which rule sent it there.

  • Reps approve stage changes

    The agent proposes next steps, stage changes and amounts from call notes, and the owner approves in one click. Nothing moves a deal to closed won on its own.

Frequently asked questions

Does HubSpot have an MCP server?

Yes. HubSpot runs an official remote MCP server with OAuth that reads and writes CRM records such as contacts, companies, deals, tickets, quotes and engagements, while marketing and content data are read-only. There is also a local developer MCP server for building apps. Claude, ChatGPT and other MCP clients can use it directly, so connecting an assistant needs configuration rather than code.

Is Breeze enough, or do I need a custom HubSpot agent?

Breeze is enough when the answers and actions live in HubSpot and you are on Pro or Enterprise. You need a custom agent when it must check another system first, such as the ERP, billing or a scheduling tool, when routing rules must be auditable in code, or when conversations happen on channels an outside agent handles, such as WhatsApp or the phone.

Can AI update HubSpot automatically after sales calls?

Yes, and it is one of the most useful builds. After each call the system logs an engagement associated with the right contact, company and deal, drafts the note and a follow-up task, and proposes property changes from what was said. The rep approves them in one step, and HubSpot's property history shows the integration as the source, so nobody wonders where a value came from.

Can I connect WhatsApp Business to HubSpot with AI?

Yes. HubSpot's own inbox can connect WhatsApp on some tiers. For an agent that qualifies leads in WhatsApp, the agent runs on the Cloud API, finds or creates the contact by phone number, logs the conversation, and opens the deal once the lead qualifies, all through HubSpot's API. Opt-in and a route to a person are required either way.

Also browse: Sales and lead handling · Agencies and professional services · Salesforce · WhatsApp · all use cases