For the past two years, almost every software company has announced an AI assistant.
Most of them have been useful.
Some have saved us a few clicks. Others have helped write emails, generate reports or answer questions buried inside our CRM.
But very few have fundamentally changed how work gets done.
HubSpot's announcement of Agent Builder and Agent Hub feels different.
Not because it introduces AI agents. We all knew they were coming. It's different because it changes where automation happens, and how modern go-to-market teams can orchestrate work across their entire technology stack.
I think this is one of the clearest signs yet that we are entering the era of agentic GTM.
We have been here before
Every major shift in software has followed a familiar pattern.
First, software helped us store information.
Then it helped us organise information.
Then it helped us automate processes.
Now, it is starting to complete work on our behalf.
That might sound like a subtle distinction, but it is a profound one.
Traditional workflow automation follows predefined rules. If X happens, do Y.
AI agents are different. They can gather information, make decisions, choose actions and even collaborate with other agents to complete more complex objectives.
That is a completely different category of software.
I think we will look back on the next few years as the period where software stopped being something we used, and became something we managed.
Why this release matters
On the surface, Agent Builder looks like another visual workflow tool.
It isn't.
The important part is not that you can build agents. It is what those agents can connect to.
For years, HubSpot Workflows have revolved around CRM events. A contact is created. A deal moves stage. A property changes. Everything starts inside HubSpot.
Agent Builder removes that restriction.
Now an agent can begin because someone adds a row to a Google Sheet. Or because your product sends a webhook. Or because an intent platform detects buying signals. Or because another application triggers an event.
That fundamentally changes HubSpot's role.
Instead of being the place where customer data lives, it becomes the platform that orchestrates work across your entire go-to-market ecosystem.
That is a much bigger ambition than building a better CRM.
AI stops being a feature
One thing I particularly like about HubSpot's approach is that AI is not presented as a destination.
It is simply another action inside a workflow.
Rather than opening ChatGPT to ask it to write an email, the workflow simply reaches out to an LLM when it needs one.
Imagine something like this:
- Website visitor shows buying intent
- Research the company
- Enrich the CRM
- Use GPT-5 to personalise outreach
- Score commercial opportunity
- Assign to the correct salesperson
- Create a LinkedIn audience
- Notify Slack
- Update reporting
Nobody has to ask the AI to do anything. The process simply happens.
That is where I think enterprise AI becomes genuinely valuable. Not because people spend more time talking to AI, but because they spend less time doing repetitive operational work.
From marketing stacks to agent stacks
For years we have obsessed over building the perfect marketing stack.
CRM. Email platform. Advertising platforms. Analytics. Sales engagement. Customer success.
The challenge has never really been the tools. It has been getting them to work together.
Agent Builder feels like a step towards solving that problem.
Instead of humans sitting between every platform, agents become the connective tissue. They gather data, interpret it, decide what should happen next, then trigger the appropriate actions.
Increasingly, I think businesses will not be talking about their software stack. They will be talking about their agent stack.
The rise of specialist agents
Whenever people talk about AI, they often imagine one super-intelligent assistant that does everything.
I do not think that is where businesses are heading.
I think they will end up with dozens, perhaps hundreds, of specialist agents. Each doing one job exceptionally well.
Imagine a growth team with agents responsible for:
- Lead qualification
- Proposal generation
- Competitor monitoring
- AI search visibility
- Digital PR opportunities
- Technical SEO auditing
- PPC optimisation
- Customer onboarding
- Executive reporting
- Renewal management
None of these agents needs to be particularly complicated. Their value comes from specialising.
Even more interestingly, HubSpot's Agent Builder allows agents to invoke other agents. That means you can start composing workflows from intelligent building blocks rather than static automation rules.
From a software architecture perspective, that is a fascinating direction.
What this means for agencies
This is the section that excites me the most.
Historically, agencies have sold deliverables. SEO. Paid search. Content. Digital PR. Email marketing.
Increasingly, I think agencies will sell operational systems instead.
Imagine onboarding a new client. Instead of manually configuring dozens of tools, importing contacts, assigning tasks and creating reports, an onboarding agent provisions everything automatically.
Imagine an AI search agency. One agent monitors AI visibility. Another analyses competitor citations. Another identifies opportunities for Digital PR. Another generates content briefs. Another produces monthly executive reports.
That is a completely different service model.
The value is not simply producing deliverables. It is designing systems that continuously generate them.
The first five agents I would build
Having spent the last year thinking about AI search, answer engine optimisation and machine-readable reputation, I already have a long list of ideas I would love to experiment with.
1. AI Search Visibility Agent. Monitor ChatGPT, Gemini, Claude and Perplexity for changes in brand visibility, citations and recommendations. Notify the team whenever competitors gain ground.
2. Citation Intelligence Agent. Track where brands are being cited across reviews, editorial coverage, forums and industry publications. Identify gaps in machine-readable reputation before they affect AI recommendations.
3. Digital PR Opportunity Agent. Monitor news, journalists and competitor activity. Recommend proactive outreach opportunities based on emerging topics.
4. AEO Audit Agent. Automatically audit new client websites for structured data, entity optimisation, reviews, citations and other factors that influence AI understanding. Generate prioritised action plans without requiring manual audits.
5. Executive Growth Agent. Pull together HubSpot, GA4, Search Console, advertising platforms and AI search visibility into a weekly board-ready summary. Highlight anomalies. Explain why they happened. Recommend what to do next.
Those are the kinds of agents that do not just save time. They create capability.
This isn't about replacing people
Every AI announcement inevitably leads to discussions about replacing jobs.
Personally, I think that is the wrong conversation.
The biggest opportunity is not replacing marketers or salespeople. It is removing the repetitive operational work that sits between strategy and execution.
Updating CRM records. Moving data between systems. Generating routine reports. Creating audiences. Assigning tasks. Checking intent platforms. Researching prospects.
Those are all things agents should be doing.
That allows people to spend more time solving problems, building relationships and making commercial decisions.
The best growth teams will not have fewer people. They will have fewer repetitive processes.
My first impressions
I have been a HubSpot user for years, and it is easy to become cynical every time another AI feature is announced.
This release feels different.
Not because it is perfect. Not because every business should rush to deploy hundreds of AI agents. But because it signals where software is going.
HubSpot is no longer thinking purely about CRM. It is thinking about orchestrating work across the entire go-to-market function.
Whether HubSpot ultimately becomes the dominant platform for agentic workflows remains to be seen. Salesforce, Microsoft, Google and plenty of startups are all racing towards the same destination.
But I think HubSpot has made a very important move.
Final thoughts
We have spent the last twenty years optimising how people interact with software.
I think the next twenty years will be about optimising how software interacts with software.
People will still provide strategy. They will still exercise judgement. They will still build relationships.
But more and more of the operational work between those activities will be delegated to intelligent systems.
HubSpot's Agent Builder is not just another AI feature. It is another signal that we are entering that world.
If you work in marketing, sales, RevOps or agency leadership, I think it is well worth paying attention to.
