HubSpot AI Agents: Automating RevOps for Marketers | MAN Digital

Most features are in beta and tie directly to native tools. They are free during beta with no credits required. This means you should test high-impact agents now before pricing begins. The real opportunity lies in integrations and custom builds beyond native capabilities.

How to prioritize:

• Start with the Customer Agent for ticket queues, the agent that eliminates your biggest time sink.
• Prospecting Agent for SDR research,
• Data Agent for property hygiene.

What is HubSpot Breeze AI?

Breeze AI automates routine tasks, learns from your data, and connects your stack to shift your focus from admin to revenue work.

Imagine 30% more selling time for reps, 50% faster lead response, and clean data without manual entry.

Why this shift matters: faster handoffs, fewer dropped balls, reliable dashboards.

It operates in two modes:

• HubSpot AI Agents function as AI employees. For example, the Prospecting Agent monitors 100 contacts monthly and drafts personalized outreach when buying signals appear.

• How: set signals (job change, intent), define sequences, and require approval before sending for the first 30 days.

• Assistants analyze data and create content, like asking 'What deals closed last quarter?' but can't send reminders.

• Use when you need insight or drafts without risk of execution.

Why two modes? Not every task should run on autopilot. Assistants remain read-only, while agents execute with guardrails.

You manage every aspect of your AI team:

• Define outcomes and KPIs, ensuring AI aligns with revenue objectives.
• Gate data access by role to protect sensitive deal information.
• Customize tone and voice to match your brand while maintaining trust with enterprise buyers.
• Build rules and guardrails to protect your process, preventing expensive automation errors.

These AI tools use your HubSpot data and connected apps while your team maintains control through reviews, approvals, and audit trails for significant actions.

Agents vs Assistants

Breeze Studio offers two distinct configuration paths:

• Assistants analyze CRM data and browse the web for insights.
• Agents read data and execute actions through tools and MCP servers.

When to use which:

• Assistant: Ask questions, such as “How many contacts came from Google Ads last week?” or draft responses.
• Agent: Enforce processes such as enriching leads, scoring opportunities, updating properties, notifying owners, or creating content.

Anatomy of an AI Agent

Every agent builds on three fundamental elements:

1. Instructions
   • The system prompt defines the role, rules, goals, tone, and guidelines.
   • Optional welcome message and prompt patterns for common tasks
2. Capabilities (what it can do)
   • Tools: Fetch data, generate content, and take action.
   • MCP servers: External services and APIs
3. Knowledge
   • Private knowledge base: brand guidelines, ideal customer profiles, FAQs, product documentation, service agreements, and company policies.
   • Limit knowledge to 3–5 core files for targeted, on-brand outputs.

The configuration formula:

• Assistants = Instructions + Knowledge + Read-only access
• Agents = Instructions + Knowledge + Action tools or MCP

How AI Agents Use Tools and MCP Servers

• Tools let agents fetch records, update properties, send messages, and generate content. Custom tools development is in beta, allowing you to create your own

• MCP servers provide secure "world connections" for your agents. They support HubSpot by connecting to Asana, Jira/Atlassian, and Notion, with additional options as developers build new MCPs. You can develop yours.

22 Native HubSpot Breeze Agents & Assistants (Beta)

HubSpot's library includes 22 AI agents and assistants you can use right away.

This overview reflects the current Breeze catalog. Plans and credits remain in beta and will change.

Why we like them: native security, faster setup, and lower maintenance than using multiple point tools.

What are the 4 main AI agents in HubSpot?

Four flagship AI agents integrate into the platform's UI and workflows, each solving specific revenue challenges. Start with the one matching your biggest time sink:

• Social Agent automates post creation, scheduling, and engagement. It uses AI-generated visuals and tailored optimization for each platform.
• Prospecting Agent monitors contacts, researches companies, and drafts personalized outreach sequences based on buying signals.
• The Customer Agent handles support conversations via chat, email, and WhatsApp, updates CRM data in real-time, and escalates issues to human agents.
• Data Agent generates insights and responses for specific records, analyzes trends, and surfaces intelligence from your CRM data.

These flagship agents provide native AI capabilities with deeper integration than marketplace apps. Additional specialized agents extend these core functions for specific use cases.

How HubSpot Credits Work

HubSpot Credits power AI features across the platform. Your monthly allocation depends on your subscription tier:

What uses credits:

• Customer Agent: 100 credits per conversation (~$1)
• Smart Properties: 10 credits per field per record
• Data Enrichment: 10 credits per enriched record
• Prospecting Agent: 10 credits per new company added
• AI Workflow Actions: 10 credits per Breeze action

Key points:

• Credits reset monthly with no carryover.
• Additional credits cost $10 per 1,000.
• Set monthly spending limits.
• Beta features are free now, but they will consume credits later. Plan to avoid unexpected charges.

Prioritize credits by impact:

• Customer Agent for support teams (immediate CSAT improvement),
• Focus on target accounts and open opportunities for data enrichment for high-value leads.
• First, test other sandbox features — validate credit burn versus outcome.
• Assign credit ownership to RevOps for control, and publish a monthly usage report.

Conclusion

Start with a small, effective AI team.

It depends on your GTM motion, but the pattern remains: start narrow, tie each agent to one KPI, and expand when the revenue impact shows. Focus prevents “AI overload” and preserves credits.

These tools don't replace your team. They eliminate the administrative burden, allowing Marketing, Sales, and Service to work more efficiently on pipeline and retention.

How we measure:

• Speed-to-lead,
• SLA adherence,
• Data completeness,
• Win rate
• CSAT (Customer Satisfaction Score)
• NRR (Net revenue retention)

Key insights:

• Keep humans central: ensure approvals for sensitive changes, provide clear guidelines, and maintain audit logs.

• Use tools for repeatability. Use MCP to transform insights into cross-app actions.

• Business metrics: speed-to-lead, resolution time, data completeness, win rate, CSAT, NRR.

Next steps (90 days):

1. Baseline (Weeks 1-2):
   • Audit data health (target: 80% contact completeness)
   • Measure SLAs (baseline: current response time)
     Identify manual tasks (find 5+ automation opportunities), and select team KPIs (pick 1 per department).
2. Pilot (Weeks 3–6): Launch targeted pilots and add success metrics to each to define success.
   • Marketing: Deploy Social Assistant for campaign enhancement — measure time saved and engagement increase.
   • Sales: Launch Prospecting Agent for contact monitoring. The agent will monitor 100 contacts, approve before sending, and track reply rate and meetings.
   • Service: Implement Customer Agent for triage support — enforce escalation rules, target FRT < 20 minutes
3. Operationalize (Weeks 7–12):
   • Add human checkpoints,
   • Publish role-based dashboards,
   • Extend MCP
4. Hold a 30-day review to assess KPIs. If they improve, add one more use case per team; if not, refine instructions, tools, or knowledge.
   • Use a simple post-mortem: what worked, what broke, what to change next sprint.

Execute well for faster handoffs, cleaner data, and shorter cycles. You’ll see an AI “bench” that compounds revenue results without adding headcount.

At MAN Digital, we're building custom agents and apps for HubSpot. If you want to discuss, we're happy to share what's working for our clients.