Every company I talk to right now is convinced they have an AI problem.
Their AI writes emails that no one responds to. Review the accounts and identify leads from the sales team that were already closed six months ago. Copy and paste sessions between tools generate content that sounds exactly like what each competitor has published. Leaders invest in tool after tool, run training session after training session, and still find themselves staring at the same question: Why isn’t AI actually moving the needle?
Here’s what they don’t tell you. The problem is not your model. The problem is not your data. The problem is context: specific knowledge about your business, your customers, what they need right now, and how your team actually works. It’s also the hardest problem to solve and the one the industry has been slowest to address.
The context is infrastructure, not functionality
Here is the distinction that I think is being lost. The data is what happened. Context provides meaning to actual events, what they mean, why they matter, and what to do about them. Context is not a feature; it is a necessary infrastructure.
Your CRM recorded that a deal closed eighteen months ago. This is the data. The context is knowing that the deal was closed because your sample changed companies, the price had to be changed three times before arriving, and that customer now reports several new deals a year and hates being contacted by automation. A human being who worked on that account knows all this. Almost no AI does this, because almost no platform is built to capture it.
This is the gap. Not a model gap. No data gaps. A context gap. And that’s the problem HubSpot is solving with the Agentic Customer Platform. When Yamini introduced our Agentic Customer platform earlier this year, she described the foundation behind it: a place where all your customer data and business context resides, available to your team and your AI agents when they need it.
The best infrastructure is invisible. It runs in the background, stays up to date as your business changes, and doesn’t force your team to iterate. This is the standard that AI is supposed to hold itself to and almost never meets.
The hidden cost of context gaps
There’s a cost your team pays every single day that doesn’t show up in your AI budget. We call it the briefing tax: the time and repetition needed to give the AI enough background to produce something useful.
You explain your brand voice before asking them to write. Paste your account history before asking them to search. Describe your pricing structure, competitive landscape, customer profile, before any significant activity. And the next day, you do it again. It doesn’t learn your business. The real cost isn’t the hours your team wastes re-briefing the AI, but the opportunity cost: the information the AI could have surfaced if it actually knew your business.
The briefing fee is just the daily friction. The hardest problem is the one you don’t see: what happens to the context over time. Your competitive positioning changes. Your ideal customer profile changes. Your playbook is updated. Your AI knows nothing about it. It’s not like he forgot about it. He remembers the conversation. It simply has no connection to the company behind it.
For GTM teams, it seems that the AI is definitely wrong. A project changes, your team adapts, but the AI continues to draw on an outdated context. The exits start ringing. The advice no longer fits your goals.
When your AI isn’t connected to the full picture, it can never develop the comprehensive, dynamic knowledge it needs to create real value. It stays A tool. He never becomes a trusted teammate.
Growth teams need their own context
Not all context is the same. Personal AI tools like ChatGPT are building personal context: your preferences, your conversation history, your communication style. Enterprise tools like Glean are building the organizational context: your documents, wikis, and institutional knowledge. At HubSpot, we’re building growth context: the rich, precise, high-quality understanding that AI needs to drive results in marketing, sales, and customer success.
This is not a concept. We are building a real infrastructure that will mean capturing and maintaining this context for customers, while also giving them the ability to self-manage. We view the growth context as having five dimensions:
- Commercial activity context it’s everything you do, how you compete, and what makes you worth buying. Your product positioning, your differentiation, your pricing logic, your brand voice. This is the context that makes AI seem like your company instead of like every other company. your category. Capturing it requires more than simply uploading a branded document. It requires a system that structures that knowledge and automatically applies it in every interaction.
- Team context this is how your employees actually work. Your sales methodology, your qualification criteria, your escalation paths. Not the version found in your onboarding documents, but the version actually used by your best reps. This is what distinguishes an AI that follows a script from an AI that exercises real judgment. This type of context does not exist in any CRM field. It lives in call recordings, chord notes, and patterns visible only in thousands of interactions.
- Context of the trial this is what your workflows look like in practice. What triggers a handover. What makes a deal high priority. How your campaigns are built and how successful each of them will be. This is what allows AI to act, not just provide information. Integrating this into AI requires understanding actual workflows, not just describing them, so that the system can act on them rather than reference them.
- Customer context it is the accumulated history of your relationships. What each account bought, why they bought it, what their goals are, where friction occurred, what the next logical conversation should be. This is what makes outreach feel like a conversation instead of a cold call. This is the hardest category to maintain because it changes constantly. Maintaining this update automatically, at every touchpoint, is the infrastructure problem that most platforms have not solved.
- Network context it is the one dimension of the growth environment that no single company can build alone. HubSpot works with more than 280,000 businesses. This means we see broad trends in how teams go to market, how campaigns perform, and how customers buy, on a scale that no single company could replicate on its own. That collective intelligence becomes a layer of growth context available to every business on the platform, shaping what your AI recommends before you’ve even launched a single campaign.
How are the right questions
If you’re evaluating AI for your team, the questions that really matter aren’t about the model. Models are increasingly commoditized. The right questions are about context.
- Can he do it? capture and act on the complete picture? Not just the structured and unstructured data in your CRM, but the reasoning, judgment, and institutional knowledge that typically resides in people’s heads.
- Is the context maintained automatically? Or does your team have to keep it updated manually, turning the investment in a platform into a maintenance burden?
- Is it built specifically for growth? Or is it a generic level of knowledge that includes some customer data?
- Does it get worse over time? Or is constant reinvestment necessary to stay relevant?
Answer “no” to any of these questions and your AI doesn’t work for your business, it works on a version of your business that it no longer exists.
This is the real AI race. Companies that get the growth context right don’t just make better use of AI. They get further every time they use it.
