OC HUG Recap: Where to Start With HubSpot AI Without Getting Overwhelmed
By
At the August Orange County HubSpot User Group, held in person in Irvine and streamed to attendees from as far as Europe and Africa, Maria from HubSpot skipped the product tour and asked a sharper question: how can HubSpot AI actually help your business work smarter? Her answer was to start with one painful, repetitive task, not every feature at once.
Where teams lose their time
The same 20 follow-ups go out every week, reps spend hours researching before calls, support answers the same questions over and over, and decisions wait while someone builds a report. Maria cited a figure that roughly 65% of a sales rep's time goes to admin rather than talking to customers. The goal for AI is to remove those tasks. The question isn't which AI app to buy; it's which repetitive task would matter most if it disappeared.
Four pieces that work together
- Breeze Assistant: an AI helper inside HubSpot on every plan, already grounded in your CRM. It drafts, summarizes and answers questions in the context of your role and customers, and it doesn't use credits.
- Breeze agents: purpose-built agents for specific jobs like prospecting, customer support and content.
- AEO (answer engine optimization): making sure your brand is found and cited in AI search tools like ChatGPT, Gemini and Perplexity.
- Loop Marketing: HubSpot's model for a continuous cycle of content, distribution and feedback, kept running by AI.
Homework from the session: click the Breeze icon in your HubSpot account and ask it to summarize your last three customer calls, or draft your next follow-up emails.
Three agents worth knowing
Data Agent
Answers questions about your CRM data without building a report, such as last month's click-through rate, and enriches records. At the meeting, John Hulina shared a client example: a Data Agent smart property that reads meeting transcripts and emails for budget mentions and fills in the budget field for a team selling with the MEDDPICC methodology.
Customer Agent
Handles routine support questions around the clock, trained on your knowledge base, past conversations and up to 1,000 website pages, and hands complex cases to a person with full context. HubSpot's customer example: Soundstripe, a music licensing platform, had the agent handling 79% of chat volume across WhatsApp, email and chat within a quarter, with faster first responses and resolutions, without cutting staff.
Prospecting Agent
Does the account research and briefing that used to take reps 15–20 minutes per call, surfaces buying signals, and drafts sequences and follow-ups for your review. HubSpot cited a software firm that runs four agents, one per customer segment, and said the work of 10 BDRs now takes three. The agent includes a 14-day free trial.
A live example: an AI agent for loan intake
When Maria asked what task attendees would hand to an agent first, John described a commercial lender client that receives a steady stream of broker loan requests, many too big, too small or the wrong type. The agent reads each email, creates the contact and lead, checks it against lending criteria, and drafts the reply. Every result is tracked in HubSpot, so the lender can see which brokers keep sending poor-fit deals and coach them on what it funds. Read how that build came together.
Start with one problem, in three steps
- Name the friction point: the one task that would have the biggest business impact if it were 50% faster. Start only there.
- Set a metric: define what working looks like, for example "30% of tickets resolved without a human in 60 days." The number matters less than having one.
- Fix the knowledge base: AI is only as good as what you feed it. If your documentation is scattered, outdated or in people's heads, that's your first project.
How credits work
AI features run on monthly credits that don't roll over: 500 on Starter, 3,000–5,000 on Professional and 5,000–10,000 on Enterprise, with extra packs at $10 per 1,000. Typical costs are 50 credits per conversation the Customer Agent resolves, 10 per Data Agent smart property run, and 100 per Prospecting Agent lead. You can set a limit so agents stop once the included credits are used. John shared a cautionary example of a portal that burned its monthly credits on the first day, which is why setup and limits matter.
AEO and Loop Marketing
Buyers now ask AI before they visit your website. HubSpot's data shows organic search traffic down about 20% year over year while traffic from AI answers has roughly tripled, and AI-referred visitors tend to convert better because they arrive more informed. HubSpot AEO tracks how often your brand appears in AI answers, benchmarks competitors and recommends fixes. It's included in Marketing Hub Professional and Enterprise, or available as an add-on.
Loop Marketing ties the tools to the buyer journey: AEO and the Content Agent to attract, Breeze Assistant and the Prospecting Agent to engage, smart deal progression to close, the Customer Agent to retain, and the Data Agent to learn and improve.
Q&A highlights
Why does Breeze get a HubSpot formula right when Claude or ChatGPT doesn't? Context. Breeze has your portal's data and knows HubSpot's syntax. John's workflow: try Breeze first, then an outside AI. He also advised using AI to help write requirements, not to design the solution.
Can HubSpot personalize outreach timing and content by region or engagement? That's the Prospecting Agent's job. Unlike marketing emails or sales sequences, it researches each contact and your website to personalize every message. Start in manual review mode and set credit limits before switching to automatic.
Where should you start? Ask Breeze first, use HubSpot's free small-group coaching sessions to set agents up correctly, then run a trial before scaling.
Join the next OC HUG
The Orange County HubSpot User Group alternates virtual and in-person meetings in Irvine. Get in touch with John Hulina to join or suggest a topic.
