Is your HubSpot ready for AI?

HubSpot's AI can make your team more productive, but only when the data underneath is ready. Most portals aren't yet, and that's fixable. I check what HubSpot's AI would find in your portal, fix the foundations, then switch on the parts that will earn their licence.

HubSpot AI readiness means your portal has what Breeze and HubSpot's AI agents need to work: records that match your business, complete and consistent data, and history brought in from other systems. I'm an independent HubSpot Gold Solutions Partner. My fixed-price readiness check tells you what's ready, what isn't and what to fix first.

How it works, in detail

Why does HubSpot's AI disappoint on so many portals?

HubSpot's assistant and agents read the same records your team does. If a deal's real value sits in a spreadsheet, the AI can't see it. If a customer's twelve sites are twelve unrelated companies, the AI can't tell you about the customer.

So when the AI disappoints, it's usually the foundations rather than the AI. Fix the data and connect the other systems, and the same Breeze features start giving useful answers. That's why AI comes last in the order I work in: data, then integrations, automation and apps, then AI.

What do Breeze and HubSpot's AI agents need underneath?

  • Records that describe the business. Customers, sites, products, contracts and bookings as proper records, linked to each other. See data and CRM architecture.
  • Data that is filled in and consistent. Dropdowns rather than free text, and the right fields required at each pipeline stage.
  • History in HubSpot. Orders, invoices, tickets and usage brought in through integrations, so the AI sees the whole account.
  • No duplicates. If one customer is two records, the AI only sees half the history on each.
  • Content to answer from. For customer-facing AI on Service Hub, that's a knowledge base with real, current articles.
  • Sensible permissions, so AI features only show people what they should see.

AI is only as good as the data under it.

Same portal, same question. Drag the slider and watch what cleaner data does to the answer.

Asked of the CRM: which deals will close this quarter?

£2.41m

across 61 deals, by the CRM's count.

ConfidenceLow
  • 14 deals counted twice
  • 9 deals with no close date
  • 22 deals with no owner
  • Every figure traced back to a record.

Illustrative figures.

10%

What does the AI readiness check cover?

It's a fixed-price, fixed-scope look at your portal through the eyes of HubSpot's AI. It's also a sensible first step if you're not sure what you need yet.

  1. A short call on what you hoped AI would do: forecasting, prospecting, customer service, reporting, or just less admin.
  2. The data model. Whether your records describe the business well enough for AI to work with.
  3. Data quality. How complete the important fields are, free text that should be dropdowns, and duplicates.
  4. Integrations. Which data AI would need that currently lives outside HubSpot.
  5. Your tier. What's available, what's switched on, and what HubSpot's AI can do for you today.
  6. The plan. Fixes in priority order with rough effort, plus the AI features worth switching on now.

The plan is yours to keep, whoever delivers it.

When should you use automation instead of AI?

A lot of what gets called AI should just be a workflow. If a job has one right answer every time, such as routing a lead by country, setting a renewal date or creating a task at a stage, automation does it the same way every time.

AI earns its place where judgement or language is involved, like summarising a long ticket history, sorting free text into categories or drafting a first reply. I'll go through your processes and tell you which is which.

What AI work looks like in practice

Breeze on a custom object. For a UK manufacturer whose test history I'd rebuilt as proper records, I wrote a Breeze prompt for logging test results and tested it in their portal. It only works because those records exist in the first place.

Sorting support tickets privately. As part of a support desk migration, I built a classifier that sorts tickets into categories using a model that keeps nothing afterwards. I checked it against tickets where the right category was already known before using it on open tickets. It means you can use AI on customer text without handing that text to a vendor that keeps it.

Claude and ChatGPT for teams. Plenty of teams already use one of them, often on personal accounts. I help pick the right business plan, set plain rules on what data goes in and train people to get good results. It sits alongside HubSpot rather than replacing it.

AI in my own delivery. I build with Claude every day, and it's a big part of why custom apps, migrations and integrations get done quickly. Everything is built and tested in a sandbox first, then released to your live portal.

Building AI-ready from day one

If you're moving to HubSpot, you don't need a readiness check later. I'll build the foundations properly during the implementation. If you already have HubSpot and the problems are wider than AI, start with a HubSpot review, from £1,000, which covers AI as one part of the whole portal.

Not sure where to start? The readiness check is the usual first step. It's a short, fixed-price piece of work that tells you whether the problem is the AI or what's underneath it.
Peter is an expert in large language models and is constantly identifying new workflows and automations to reduce manual processes and data management that helped our company immensely.
John RankinChief Operating Officer, Beonic, ASX-listed global technology business

Questions people ask

What is HubSpot Breeze?

Breeze is the name HubSpot uses for the AI built into the platform, including its assistant and its AI agents. It works from the records and content in your portal. HubSpot changes the features and names often, so I check what's current for your tier during the readiness check.

Why does HubSpot's AI give poor answers on our portal?

Usually because of what it's reading. Free-text fields, missing values, duplicates and history stuck in other systems all give AI less to go on. Fix those and the same features give much better answers.

Should we use AI or automation?

If a job has one right answer every time, use a workflow, because it does it the same way every time and AI doesn't. Use AI where judgement or language is involved, such as summarising, sorting or drafting.

Can we use Claude or ChatGPT with our HubSpot data?

Yes, with care about which plan you're on and where the data goes. I help teams pick the right plan, set simple rules on what goes in, and train people to use it well. HubSpot's own AI stays the main story inside the CRM.

Is our customer data safe?

Client data stays in client systems wherever possible. Where text has to go through an AI model, I use set-ups where nothing is kept afterwards, and business plans that come with a proper data agreement.

Where to next

Want to know if your portal is ready for AI?

Message me or book a short call. I'll ask how your portal is set up and what you hoped AI would do, and tell you whether a readiness check is worth it.

Book a call