[Q&A] HubSpot’s New AI Features, Explained by the Project Lead

At the September 2025 INBOUND Spotlight, HubSpot dropped a bomb – actually, a few hundred bombs. In mere hours, the software platform giant announced 200+ product updates and features, a Loop Marketing AI playbook for customers, and an AI Ready Partner Playbook for agencies.

What does this all mean for your day-to-day? Or your marketing agency’s?

To give you answers for 1, 3, and 5 years from now, I sat down with Justin Graci, manager of Partner GTM and enablement at HubSpot. We discussed the evolving customer-agency relationship and model, the impact of AI, and what’s next for HubSpot partners after these huge product releases.

Q&A: HubSpot’s New AI Features w/ Justin Graci

(Interview is lightly edited for clarity and brevity.)

Q: So. AI. Hot topic. There's a ton of digital marketing agencies here. Since you can tell the future, what does an agency model look like, 1, 3, 5 years out?

Justin: "We just released the AI Ready Partner Playbook, which I’ve been working on for months. It outlines how agencies can evolve in this moment. I’ve spoken with hundreds of partners over the past few years, and the biggest takeaway is that the model of the future is leaner, more efficient, and higher up the value chain.

We’re already seeing AI-native consulting firms with just five employees competing against traditional agencies of 30, 40, or more. AI automates commoditized work, so agencies need to move into higher-value services. The firms that thrive will be the ones who evolve their offerings, streamline operations, and adopt AI not to replace people, but to amplify them."

I attended the agency blueprint yesterday on AI. People that might historically be developers can do other things because AI can support something that maybe they weren't comfortable with before.

Justin: Exactly. One partner in the UK built an onboarding agent to handle InfoSec questionnaires, which used to require manual review. They automated the process, saving each employee 200 hours a year – about $200,000 in value.

What’s key is they didn’t cut staff. Instead, they reassigned employees to higher-value work. That’s the opportunity here: AI takes the repetitive, time-consuming tasks off the table, freeing teams to do more impactful work.

Q: There are so many AI tools, and there are going to be more and more. How does an agency hone in?

Justin: What we're seeing right now is the same thing that happened with SaaS with point solutions, and this was actually what started the whole ‘crafted, not cobbled’ narrative that HubSpot started pushing a few years ago. Why buy all these different tools that aren't connected?

They're all fragmented. They don't talk to each other, or to make them talk to each other, you need to integrate them properly, which could cost a lot of money vs. an out-of-the-box platform that’s using the same context, the same data. 

The way I see it is that soon, everything's going to start consolidating, and the platform companies will start to really eat the cake here. So I think HubSpot's well positioned there, having best-in-class AI functionality in our platform and then best-of-breed integrations that work seamlessly with our platform. That's where we have the connector to ChatGPT, to Claude, and now Gemini we announced today.

And those work perfectly in sync with HubSpot with being able to pull the context and the data out of your CRM and then push it back. That's what's going to win. But to answer your question about how we help end users focus … every team, marketing, sales, service, they're all trying all these different tools, whether it's company-approved or not. Almost every AI company product that I've seen, they all have a freemium version or a fairly robust trial version. 

We need to help end users start to focus and narrow and say, ‘Here's the value you can get from a platform.’ What's going to happen when all these (one-off) tools don't work in a connected workflow? Also, eventually the cost of all those tools will add up. The security and governance around the use of those tools also becomes an issue. So those are the things you have to point out to an end user.

But I think you also have to allow for a little bit of the experimentation and playfulness that's going to happen because we're in that stage where no one knows yet. But then soon we're going to have to reel it in. And I think that's where HubSpot can come in. With the latest updates, Breeze Assistant can offer ChatGPT-level responses and keeps memory. It's gotten way more robust.

The new addition of memory is awesome.

Justin: Yeah. Memory and then the custom assistant stuff. You'll be able to now customize your assistants from the ground up, leveraging all of the (language learning models) that we have. That was the reason why I think a lot of people weren't using Breeze Assistant originally. It lacked the robustness that it needed and the context and the memory and all that. And that's why people went and used ChatGPT directly, and I think more and more, even myself, will start to use (HubSpot’s) platform tools as it gets more powerful.

Q: We've been deploying HubSpot for many years. I can't tell you how many times I've been asked, ‘Can you build this so I can have a project management tool?’ How close are we to that with the new Project object feature and Gantt chart?

Justin: The Project object is one piece. The Gantt chart view is another product feature that is in beta right now.

I was actually just playing around with it earlier, and it's pretty slick. It's early on with these products. I think come next spring, we're going to see it really start to come to life.

Let's say whoever is selling deals, whether it's a salesperson or account manager, they close a deal for HubSpot, and then you need to kick off a service project for the onboarding and implementation for the customer. So you can basically have the rep associated as the owner to the deal.

And then the deal, once that is closed, it spins up the Project object automatically that has all of the preconfigured tasks (and soon) subtasks and also the service lead owners on the project. So then you almost get that seamless handoff: deal close, project kicks off, onboarding happens, here's the task. And now I can see who's assigned, when it's going to kick off, and it's all preconfigured. All without leaving HubSpot.

Q: One more question. Maybe the big one: Loop marketing.

Justin: Yeah. What do you think about it before I go into it?

We have an agency that has 25 people. There's been a lot of discussion around AI and what happens to those people. And we've always seen it as, like, those people aren't going away. But there needs to be a model, and a tool that supports that model, to help them change to the current dynamic. I love the idea of failing fast and iterating. That's what excites me. Now, convincing our clients to fail fast …

Justin: That's the challenge. But I would almost reframe it as not failing fast, but more about continuous improvement and optimization in real-time. I'm viewing Loop as the evolution of inbound (marketing). People are still going to seek out information, just in a different way.

They're going to come to your website, possibly through different channels now. So, inbound as a concept is not going away. But it's just the tactics, the channels, and the rich data that we have that we're now able to leverage. And the dream for every marketer has always been personalization at scale, but we didn’t have the data or tools to pull it off.

And now we can actually do that. With Loop, the whole Express stage is that groundwork of defining your ideal customer profile, doing all of the research, pulling in all of the rich data that you have, and really starting to craft what your story is and what's your differentiation. And then going into the Tailor stage of how do I actually now take that and personalize it specifically not just for a 100 people or 10 people, but for 1:1 personalization at scale?

I'm not alone with this, but I hate that it happens: With Instagram, I'll be scrolling, not shopping. And all of a sudden, I just get an ad for a product, usually clothing. And it's something that I didn't know I wanted or didn't know I needed, but I look at it and I'm like, ‘That is literally my style,’ or even a style that I didn't yet know I was looking for. And that level of personalization where it actually understands me and what I want, what resonates with me, and then I actually go and buy – I've done that so many times now. That's what this is going to start to unlock for B2B.

Now let me speak to the agency concern about being made obsolete by AI. With inbound marketing, anyone could have gone and wrote a blog post or a content offer and tried to optimize SEO, but they just didn't have the time. They sought out agencies to help them with it. And I think the same thing's going to happen with Loop, at least the setup and the implementation of it. Customers aren't going to know how to get started with it. They need a partner to come in and help with that. The creatives or whoever at your agency doing the work throughout the process with the client, they still have a place at each of those stages.

Someone still needs to go to define the ideal customer profile. Someone still needs to really build out that brand voice for that customer. They might be leveraging more of our AI tooling to do it more efficiently and effectively, but the customer's not going to know how to do that necessarily.

Manufacturing customers. They don't wanna do that.

Yeah. Definitely are not going to be doing it. So I think the role of people will shift a little bit, but it's no different than what they were doing before. Just different tools, more efficient and more effective with what they're doing. And I think there will be certain tasks that go away, which is actually a good thing.

Should HubSpot’s AI Features Become Your AI Features?

When you’re discovering 200+ tools and an all-new playbook in the span of an hour, it’s a lot for an agency to digest, let alone in-house marketing and sales teams. Let’s try to distill the impact Graci is predicting for 2026-2030:

Of course, no AI is failproof, especially if you use it poorly or lazily. Are you looking to deploy AI in a smart and safe way at your business? See our free, huge resource: