Something significant is happening to company structures right now. Roles are being consolidated. Teams are being asked to do more with less. Leadership are pushing hard for AI adoption, often without a clear picture of what that looks like operationally. If you're in marketing, Revenue or anywhere near a CRM, you've felt the pressure.
It's real pressure and you know what, some of it is warranted. AI genuinely is changing how go-to-market teams operate and HubSpot is moving faster than most people realise. However, the gap between where leadership thinks AI is taking the business and where the business actually is right now is significant. McKinsey's 2025 State of AI report found that while 88% of organisations now use AI in at least one business function, nearly two-thirds have not yet begun scaling it across the enterprise. That's not a slow-movers problem. That's a readiness problem, and it shows up most clearly in the CRM.
HubSpot's direction is no longer just about adding features. Since INBOUND 2024 and through its 2026 updates, HubSpot has been building toward what it calls an agentic CRM: a coordinated system where AI agents, data, and workflows operate together inside the platform to drive outcomes across marketing, sales, and service.
In practice, this means agents that do actual work. The Prospecting Agent monitors your CRM for buying signals, researches accounts, and drafts personalised outreach. The Customer Agent handles support conversations across multiple channels and resolves tickets without a human in the loop. The Data Agent answers questions about your pipeline and enriches contact records automatically. These aren't features you turn on and forget about. They're systems that run on your data, your processes, and your documented understanding of how your business works.
That last part is where most organisations run into trouble.
Here's the thing about agentic AI: it's only as good as what it's working with. An agent that enriches bad data doesn't save time, it scales mistakes. An agent that qualifies leads based on an outdated lifecycle stage definition doesn't improve conversion, it muddies the pipeline. And an agent handed an undocumented process doesn't fill in the gaps with good judgement. It fails, or worse, it confidently does the wrong thing.
McKinsey found that nearly 80% of organisations are layering AI on top of existing processes without rethinking how work actually flows. That's the core issue. Most HubSpot portals weren't built for an agentic world. They were built incrementally, by different team members, over several years, with varying levels of CRM discipline. Properties drift. Lifecycle stages get used inconsistently. Contact records go stale. The workflows that made sense two years ago haven't kept up with how the business has changed.
There's also a gap that doesn't get talked about enough: the distance between what leadership is planning and what teams are actually doing day to day. Gallup's 2025 research found that managers are twice as likely as individual contributors to report using AI frequently. BCG's 2025 global AI at Work survey found that only about one quarter of frontline employees say they receive strong leadership support for AI, and that when that support is present, positive sentiment about AI rises from 15% to 55%. The mismatch matters because AI rollouts planned at the top often land without the context of how work is actually done at ground level. The result is adoption that looks like compliance but doesn't move anything.
This isn't a people problem. It's a communication and sequencing problem. And it's more common than most leadership teams want to admit.
None of this means waiting. It means sequencing correctly.
1. The first priority is data.
Before any agent touches your CRM, you need clean contacts, accurate and consistently applied lifecycle stages, and property fields that mean the same thing to every person on your team. This isn't glamorous work, but it is the non-negotiable foundation. Agents run on what's in the system. If what's in the system is unreliable, the agents will be too.
2. The second priority is documentation.
Not high-level strategy documents, but role-level process clarity. Who does what, at what trigger, with what information, to what standard. If you can't write that down clearly, an agent can't follow it. The organisations that get the most from HubSpot's AI over the next 12 to 24 months will be the ones that have done this work first, mapping their business processes in enough detail that handing parts of them to an agent is a natural next step rather than a risk.
3. The third priority is being realistic about which workflows are actually suited to AI.
High-volume, repetitive, rules-based work is where agents perform well. Lead enrichment, first-response to support queries, data syncing, outreach sequencing for well-defined segments. Complex, relationship-dependent, or judgement-heavy work still needs humans. Even within HubSpot's own partner community, it's well-documented that the Prospecting Agent is only effective when the CRM is exceptionally clean and well-tagged, and that automated responses aren't ready for nuanced scenarios without human review. Setting honest KPIs and policies by department, rather than rolling out a blanket AI mandate, is how you get genuine performance gains rather than a lot of activity with little measurable outcome.
The teams that benefit most from HubSpot's AI push won't be the ones who turned everything on fastest. They'll be the ones who used this moment to fix what was already broken: the data hygiene they'd been deferring, the process documentation that only lived in someone's head, the CRM structure that reflected the business as it was two years ago rather than how it operates now.
As an Elite HubSpot partner with experience across complex implementations and custom integrations, we work with organisations at exactly this junction: building the data architecture and process foundations that make AI-driven operations possible, not just theoretical.
If you're trying to work out what to actually do first, that's where to start.
Talk to us about your HubSpot implementation today.
What do I need to do before using HubSpot's AI agents?
Before activating HubSpot's agents, focus on three things: clean and consistently maintained CRM data, clearly documented processes at a role level, and a realistic assessment of which workflows are suited to automation. Agents perform well on high-volume repetitive tasks with clear rules. They need reliable data and clear instructions to work effectively.
What is HubSpot Breeze and what can it actually do right now?
Breeze is HubSpot's AI layer, made up of the Breeze Assistant (an in-platform AI companion), Breeze Agents (autonomous tools that handle specific workflows), and Breeze Intelligence (data enrichment and intent scoring). Agents in general availability include the Customer Agent for support, the Prospecting Agent for sales outreach, and the Data Agent for CRM research and enrichment. Results depend heavily on the quality of your underlying CRM data and the clarity of your processes.
Why does CRM data quality matter so much for AI?
AI agents act on the information in your CRM. If contact records are incomplete, lifecycle stages are applied inconsistently, or property fields mean different things to different team members, agents will make decisions based on that unreliable information and scale those errors across your database. Data quality isn't a nice-to-have before AI adoption. It's the prerequisite.
How do I know which HubSpot workflows are ready for AI automation? Start by asking whether a human could follow the process from a written description alone. If the answer is yes, and the work is high-volume and repetitive, it's likely a good candidate for an agent. If the process relies on relationship context, personal judgement, or decisions that aren't clearly documented, keep a human in the loop. Most organisations will have a mix of both, and treating them differently is how you get real results rather than just automation for its own sake.