HubSpot just got a lot more capable. Here's what's worth using.

Last week, I spoke with two GTM leaders who said their teams are drowning in AI tools.

They’ve got Claude open in one tab, ChatGPT in another, a new workflow automation tool their VP just approved, and somewhere in the background, HubSpot.. the platform their entire business runs on.. quietly doing things they haven’t noticed yet.

Here’s what I keep coming back to: before you bring in anything new, you should know what you already have.

And right now, if you’re on HubSpot, you have more than most people realize.

When something feels painful or manual, the instinct is to go find a tool that fixes it.

I get it. That’s how most of us are wired. Problem surfaces, we start Googling, we end up in a 14-day trial for something we’re going to forget about in three weeks.

But here’s the pattern I’ve watched play out at companies of every size: they stack tools on top of problems that HubSpot could have solved natively.. and then wonder why their tech stack feels bloated, their data is scattered, and their team refuses to update the CRM.

Every tool you add outside of HubSpot is a place where data goes to die.

That’s the real cost. Not the subscription fee. The data fragmentation.

HubSpot has been quietly building AI into the core of the platform under their Breeze umbrella. Some of it is genuinely good. Some of it still needs work. Here’s my honest read on what’s worth using today.

AI workflow builder. This is the one I point to first when teams tell me they’re intimidated by automation. You describe in plain language what you want to happen.. “when a ticket is created and the customer hasn’t responded in 48 hours, send a follow-up email and notify the CSM in Slack”.. and HubSpot builds the workflow structure for you. You still review it, clean it up, and own the logic. But the blank page problem is gone. For teams that have been putting off automation because it felt too technical, this removes the biggest barrier. You don’t need a RevOps hire to get your first automations running.

Meeting notetaker. HubSpot now has a native meeting transcription tool that logs notes directly into the CRM. I’ve been watching teams pay for Gong, Fireflies or Otter on top of HubSpot when this capability now exists natively. It’s not Gong.. it doesn’t have the same depth of conversation intelligence. But for most post-sales teams that just need transcripts tied to contact and deal records without adding another integration, it does the job. Less tooling. Cleaner data. One less thing to sync.

AI email personalization and send time optimization. HubSpot can now generate personalized subject lines and email copy using your CRM data, and predict optimal send times based on individual contact behavior.. not population averages. This is genuinely useful for CS teams sending renewal sequences, onboarding communications, or QBR follow-ups. The output still needs a human pass. But the first draft being populated from your own data is faster and more relevant than starting from scratch.

Engagement scoring. HubSpot now surfaces engagement signals.. high, medium, low.. based on email opens, clicks, form fills, and website behavior. For a CSM managing 40+ accounts, this is a useful triage layer. You’re not guessing who’s going cold. You’re working from signal. This is the kind of thing teams used to pay for separately or build custom health score frameworks around.. and it now exists in the platform without configuration.

AI report and dashboard builder. Describe the report you want in plain language and HubSpot generates it. This isn’t perfect and the results need checking.. but for teams where RevOps or data resources are thin, it removes the “I need a custom report but nobody has time to build it” problem that slows down visibility.

I said I’d be honest, so here’s the other side.

HubSpot’s AI content creation tools.. auto-generated blog posts, landing page copy, social content.. are not something I’d let run unsupervised. The brand voice is inconsistent, the output is generic, and the risk of publishing something that sounds nothing like you is real. These tools are co-pilots, not authors. Treat them accordingly.

The Prospecting Agent is only effective if your CRM is exceptionally clean and well-tagged. If your data hygiene is poor, the output will reflect that. Garbage in, garbage out.. no matter how sophisticated the AI layer is.

And anything that requires deep contextual understanding of a specific customer relationship, nuanced communication, or strategic judgment? That’s still a human job. HubSpot doesn’t know your customer the way your CSM does.

Here’s my simple framework for deciding.

Use HubSpot natively when: The task lives inside a workflow, report, email sequence, or automation that touches CRM data. If the action needs to be logged, tracked, or triggered by something in HubSpot.. do it in HubSpot. Keep the data where your team already works.

Use another AI tool when: The task requires synthesis, judgment, or context that lives outside of HubSpot. Drafting a nuanced customer communication. Analyzing patterns across a set of call notes. Building a strategic framework for a QBR. Researching how to solve a problem you’ve never solved before. The ChatGPT and Claude’s of the world work best when you need to think through something complex.. not when you need to automate something repeatable.

Bring in a new tool when: You’ve confirmed HubSpot can’t do it natively, Claude can’t solve it with context alone, and the gap is painful enough to justify a new data connection, a new login, and a new thing your team has to learn. That bar should be high. Most of the time, if you dig into what HubSpot and ChatGPT/Claude can do together, you don’t need the third thing.

Before you open a new tool trial, ask yourself one question:

Does this tool create a new source of truth, or does it strengthen the one I already have?

If the answer is “new source of truth”.. pause. You’re about to make your data problem worse, not better.

The best AI setups I’ve seen are the ones where HubSpot (or whatever CRM you’re using) is the anchor, Claude is the thinking layer, and every new tool earns its place by feeding data back into the source of truth rather than creating a parallel universe of it.

This looks simple on paper. Implementation is where it actually gets complicated.. and where most teams either get it right or end up with a tool graveyard and a CRM nobody trusts.