How to Build an AI Agent for HubSpot and Slack

The workflow problem

Marketing teams generate leads. Sales teams work those leads. But the handoff between HubSpot and the sales team living in Slack is where leads go to die.

A visitor fills out a demo request form at 2:14 PM. HubSpot scores them as a Marketing Qualified Lead based on their company size, page visits, and content downloads. The lead sits in HubSpot’s MQL list. An SDR checks that list at 4:30 PM and sees the lead for the first time – two hours after a buyer raised their hand. Research shows that responding to an inbound lead within five minutes makes you nine times more likely to convert them. Two hours might as well be two days.

The problem extends beyond lead alerts. When a deal moves from “Demo Scheduled” to “Proposal Sent” in HubSpot, the marketing team that generated the lead has no visibility. They cannot see which campaigns are producing deals that actually advance, so they keep optimizing for volume rather than quality. The sales manager sees deal progression in HubSpot dashboards but checks them once a day, missing the real-time coaching opportunities that come from knowing a rep just sent a proposal or had a deal slip backward.

The communication gap between HubSpot and Slack creates delayed responses, poor handoffs, and disconnected teams that should be operating as a single revenue engine.

Why an AI agent, not just automation

HubSpot’s native Slack integration can send notifications when workflow triggers fire. But these notifications are static templates. “New MQL: John Smith from Acme Corp” tells you a lead exists but tells you nothing about why you should care about this particular lead right now.

An AI agent enriches every notification with context and judgment. When it alerts the SDR team about a new MQL, it does not just relay the form submission. It looks at the contact’s full history in HubSpot – which pages they visited, which emails they opened, which content they downloaded, how many times they have been to the site this week – and synthesizes a briefing that helps the SDR have a relevant first conversation. It might note, “This contact visited the pricing page three times this week and downloaded the Enterprise Security whitepaper. They are likely evaluating for a security-conscious enterprise deployment.”

The agent also handles the analytical side. Marketing managers can ask the agent in Slack, “How did last week’s webinar perform compared to the product launch event?” and get a comparison of registration rates, attendance, MQL conversion, and pipeline generated – without opening HubSpot, navigating to reports, and building a custom comparison. This kind of ad-hoc analysis in natural language is impossible with webhook-based automation.

How it works with ACP

The Agentic Control Plane connects to HubSpot via OAuth 2.0 and to Slack via bot token, giving the agent scoped access to both platforms.

On the HubSpot side, the agent can read and write contacts, companies, deals, and engagement data. It can access marketing event data (email sends, opens, clicks, form submissions), list memberships, lifecycle stage changes, and deal pipeline properties. On the Slack side, the agent posts messages to channels, reads messages in channels it has joined, sends direct messages, and responds to mentions.

You configure the agent with natural-language instructions. For example: “When a contact becomes an MQL in HubSpot, post a briefing to #inbound-leads with their company, role, lead score, recent activity, and a suggested talk track. When a deal changes stage, post to #revenue-updates. Answer marketing performance questions in #marketing-analytics.”

Here is how this works in practice:

Agent (in #inbound-leads): New MQL alert – Rachel Torres, VP of Engineering at DataStream Analytics (Series B, 120 employees). Lead score: 82. Recent activity: visited pricing page 4 times, downloaded “API Integration Guide,” attended Tuesday’s webinar on workflow automation. First touch: Google Ads (keyword: “api integration platform”). Suggested approach: Rachel is technically oriented and interested in integrations. Lead with API capabilities and the developer documentation. @jessica-sdrs this lead is in your territory.

Agent (in #revenue-updates): Deal stage change – DataStream Analytics - Professional Plan moved from Discovery to Demo Scheduled. Deal value: $36,000/yr. Owner: @mark-ae. Demo scheduled for Thursday at 2 PM. This deal is moving 40% faster than your average professional-tier deal. Marketing attribution: Google Ads campaign “Q1-API-Integration.”

Marketing Manager (in #marketing-analytics): @Agent how are our LinkedIn campaigns performing this month versus last month?

Agent: LinkedIn campaign comparison (March vs February): Impressions up 22% (45K vs 37K). Click-through rate steady at 1.8%. Form submissions up 31% (89 vs 68). MQLs generated: 34 vs 22 (+55%). Cost per MQL down from $142 to $118. Top-performing ad: “Secure API Gateway” creative, which generated 14 MQLs at $86 each.

The agent transforms raw HubSpot data into actionable intelligence delivered where teams already work.

Governance and security

HubSpot contains the complete history of every lead and customer interaction – emails, form data, company information, deal values, and potentially sensitive personal information. Connecting it to Slack requires robust access controls.

The ACP uses HubSpot’s OAuth 2.0 flow with granular scopes. You select exactly which HubSpot scopes the agent needs: crm.objects.contacts.read for reading contact data, crm.objects.deals.read for deal information, and so on. The agent does not receive blanket access to your HubSpot portal. If you only want it to read contacts and deals but not modify them, you grant only read scopes.

On the Slack side, the agent operates with bot-level permissions. It can only post to and read from channels it has been added to. The #inbound-leads channel might be restricted to the SDR team, while #revenue-updates is visible to sales leadership. The agent respects these channel boundaries – it posts to each channel based on your configuration but cannot access channels where it has not been invited.

The ACP audit log tracks every HubSpot API call the agent makes and every Slack message it posts. You can filter by time range, contact, deal, or action type. If a compliance review asks “what customer data was shared in Slack channels last quarter,” the audit log provides the answer in seconds.

Personally identifiable information can be filtered. You can instruct the agent to exclude phone numbers, email addresses, or other PII from Slack notifications, posting only the information that the team needs for their workflow. This helps maintain GDPR compliance when marketing data is being surfaced in a messaging platform.

Example use cases

• Instant MQL alerts with context: The moment a contact reaches MQL status in HubSpot, the agent posts a detailed briefing to the SDR channel including lead score, engagement history, content interactions, and a suggested approach. Response times drop from hours to minutes.

• Deal stage change notifications: When deals move forward or backward in the pipeline, the agent notifies the relevant channel with deal value, owner, timeline, and context about how this deal’s velocity compares to benchmarks. Sales managers can coach in real time.

• Campaign performance on demand: Marketing team members query the agent in Slack for campaign metrics – email open rates, landing page conversions, MQL counts by source, cost per acquisition – without opening HubSpot or building reports.

• Lead re-engagement alerts: The agent identifies leads that were previously active but have gone dormant (no page visits, no email opens for 30+ days) and suggests re-engagement actions to the assigned rep.

• Weekly marketing digest: Every Monday morning, the agent posts a summary of the previous week’s marketing performance to a leadership channel: total new leads, MQLs, SQLs, pipeline generated, and top-performing campaigns.

Getting started

Connect HubSpot and Slack through an AI agent in minutes.

1. Sign up at cloud.agenticcontrolplane.com. Your free account includes the connectors for both HubSpot and Slack.

2. Connect your tools. Authorize HubSpot (you will need admin or super admin access to your portal) and Slack (select your workspace and approve bot permissions) from the Data Sources page.

3. Describe your agent. Tell it your notification rules and which channels to use: “Post MQL alerts to #inbound-leads, deal changes to #revenue-updates, and answer marketing questions in #marketing-analytics.” The agent starts working immediately.