Yes — you can absolutely use AI agents with HubSpot, and you have two clear paths to do it. HubSpot ships a native set of AI agents under the Breeze brand, and it also lets you connect third-party agents through its API and OAuth. In short, you can either buy HubSpot’s built-in agents that work off your CRM data out of the box, or build/bring your own agents that plug into HubSpot as the system of record.
This guide walks you through both paths — what each native agent actually does, how to set them up, what they cost in terms of plan requirements, and where their real limitations live.
Two Paths to HubSpot AI Agents
Before you dive into setup, understand the fork in the road. The way you integrate AI agents with HubSpot falls into one of two buckets:
- Native Breeze agents — HubSpot’s own AI agents, prebuilt and running on your CRM data. No code, minimal setup, but limited to what HubSpot ships.
- Third-party or custom agents — External AI agents (built in a no-code builder, a framework like LangGraph, or a platform like Claude) that connect to HubSpot through its API and OAuth scopes. More control, more engineering.
Most teams start with native Breeze agents because they’re fast to turn on. Teams with unique workflows graduate to custom agents once the native options hit a ceiling.
What Are HubSpot’s Native Breeze Agents?
HubSpot’s AI suite is called Breeze, and it’s divided into three layers: Copilot (an in-app assistant), Agents (AI that carries out tasks autonomously), and Intelligence (data enrichment and buyer-intent signals). The agents are the part that actually do work for you.
“HubSpot’s Agent Hub is the central home for every AI agent your business runs, built directly on your CRM data.”
Here’s a plain-language breakdown of the core agents:
| Agent | What It Does | Who It’s For |
|---|---|---|
| Breeze Copilot | In-app assistant that drafts copy and answers questions about your records | Everyone — a general helper |
| Prospecting Agent | Researches leads, detects buyer intent, and drafts personalized outreach | Sales teams |
| Customer Agent | Resolves support tickets using your knowledge base, often without a human | Support teams |
| Content Agent | Generates landing pages, blogs, and social content in your brand voice | Marketing teams |
| Social Agent | Creates and schedules social posts based on past performance | Marketing/social teams |
Beyond these, the Breeze Marketplace offers more specialized agents — like a Deal Loss Agent, Customer Health Agent, and RFP Agent — that you can add for specific jobs.
Agent Hub and Agent Builder
In July 2026, HubSpot moved Agent Hub (formerly “Breeze Agents”) and Agent Builder into public beta for Professional and Enterprise customers. Agent Hub is the single dashboard where you see every agent running across your go-to-market, check their performance, and turn on the ones you haven’t activated yet.
Agent Builder is the no-code canvas where you create custom agents using your own prompts, knowledge sources, and live CRM data. You don’t need to write code — and starting September 2026, you can even describe the outcome you want in plain English and let Breeze Assistant configure the agent’s goals, tools, and data sources for you.
Native vs. Third-Party: Which Should You Choose?
This is the decision that determines everything else about your integration.
| Dimension | Native Breeze Agents | Third-Party / Custom Agents |
|---|---|---|
| Setup time | Minutes to hours | Days to weeks |
| Code required | None | Usually yes (API, OAuth scopes) |
| Data access | Your CRM data automatically | Via HubSpot API — you control scope |
| Customization | Limited to HubSpot’s templates | Full control over logic and behavior |
| Cost | Bundled with Professional/Enterprise + usage | Build cost + API/usage fees |
| Best for | Standard sales, support, and content workflows | Unique or complex business processes |
The honest answer is that most teams don’t need to choose. They run native agents for the 80% of work that’s standard, then build one or two custom agents for the edge cases HubSpot doesn’t cover.
Prerequisites: What You Need Before You Start
AI agents in HubSpot aren’t available on every plan. Here’s what you need:
- Professional or Enterprise plan — Agent Hub, Agent Builder, and the full agent suite are gated to these tiers.
- Super Admin access — You’ll need Super Admin permissions to enable agents and manage AI settings.
- Clean CRM data — How well any of this works depends almost entirely on the quality of your CRM data. Garbage in, garbage out.
- Knowledge sources — For the Customer Agent, you’ll need a populated knowledge base (help articles, FAQs) for it to draw from.
A note worth flagging: third-party reporting suggests up to 40% of AI agent deployments fail, usually because teams skip change management, ignore data quality, or deploy without clear guardrails. The tool isn’t the risk — the setup discipline is.
How to Set Up HubSpot AI Agents: A Step-by-Step Checklist
- Confirm your plan and permissions. Make sure you’re on Professional or Enterprise and that you have Super Admin access.
- Open your AI Settings. Navigate to your account settings and locate the AI settings area to review what’s available.
- Enable Breeze agents. Turn on the agents you want (Prospecting, Customer, Content, or Social) from the AI settings or Agent Hub.
- Connect your knowledge sources. For the Customer Agent, point it at your knowledge base articles and FAQs so it has accurate answers.
- Review AI trust controls. Check model cards and data-usage settings so you understand what data the agents touch and how it’s used.
- Set guardrails and approvals. Configure which actions agents can take automatically versus which require human approval.
- Test in a sandbox or small scope. Run one agent on a limited set of records or a single channel before going wide.
- Monitor performance in Agent Hub. Track what’s active, review outcomes, and iterate on prompts and knowledge sources.
Real Limitations You Should Know
AI agents in HubSpot are powerful, but they’re not magic. Here are the limitations that matter most:
- Plan gating. The full agent suite and Agent Hub are locked behind Professional and Enterprise. Starter and free tiers get very little.
- Data quality dependency. A Prospecting Agent drafting outreach off stale or duplicate contact records will produce embarrassing, wrong messages.
- Not fully autonomous (yet). Most agents still work best with human approval loops, especially for anything customer-facing or revenue-sensitive.
- Custom agents require engineering. If you outgrow the native templates, you’ll need someone who understands the HubSpot API and OAuth scopes.
- Outcome-based pricing. Some agents are priced per resolved outcome (like per ticket resolved), which means costs scale with usage — worth modeling before you scale up.
Frequently Asked Questions
Can I use AI agents with HubSpot on a Starter plan?
No. The full Breeze agent suite, Agent Hub, and Agent Builder require a Professional or Enterprise plan. Starter and free tiers get limited access to Breeze Copilot at best.
Do HubSpot AI agents require coding?
No — the native Breeze agents and the no-code Agent Builder don’t require any code. Only third-party or fully custom integrations require working with the HubSpot API and OAuth scopes.
What’s the difference between Breeze Copilot and Breeze Agents?
Breeze Copilot is an assistant that answers questions and drafts content with you. Breeze Agents are AI that carry out tasks for you autonomously — like resolving tickets or sending outreach.
Can I connect my own custom AI agent to HubSpot?
Yes. HubSpot’s API and OAuth allow you to build or connect third-party agents that read and write CRM data. This is the right path when native agents don’t cover a unique workflow.
Do HubSpot AI agents work with my existing knowledge base?
Yes. The Customer Agent, in particular, is designed to resolve support tickets using your existing knowledge base articles and FAQs as its knowledge sources.
Next Steps
Start with the low-hanging fruit. Turn on Breeze Copilot for your whole team, then enable one agent that maps to your biggest bottleneck — Prospecting if leads are slipping, Customer Agent if tickets are piling up, or Content Agent if your output is thin.
Before you scale, do the unglamorous work: clean your CRM data, connect your knowledge sources, and put approval guardrails in place. That discipline — not the AI itself — is what separates teams that see real ROI from the 40% whose deployments quietly fail.
Want to go deeper? I teach business owners how to implement AI agents step-by-step at aitokenlabs.com/aiagentmastery
About the Author
Anthony Odole is a former IBM Senior Managing Consultant, where he served as Enterprise Architect on Fortune 500 engagements, and the founder of AIToken Labs. He helps business owners cut through AI hype by focusing on practical systems that solve real operational problems.
His flagship platform, EmployAIQ, is an AI Workforce platform that enables businesses to design, train, and deploy AI Employees — AI agents that function as digital workforce members — that perform real work without adding headcount.
