Every company I talk to right now is convinced they have an AI problem.
Their AI writes emails and no one replies. Researching accounts and locations leads the sales team already closed six months ago. Fingertips of copying and pasting between tools produce content that sounds exactly like what each competitor is publishing. 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 is what you can be told. The problem is not your model. The problem is not your data. The problem is context: specific knowledge of your business, your customers and what they need right now, and how your team actually works. It’s also a very difficult problem to solve, and one that the industry has been slow to address.
Context is Infrastructure, Not a Feature
Here is a distinction that I think is being lost. Data is what happened. Context provides the meaning surrounding the actual events, what they mean, why they are important, and what should be done about them. Context is not a feature; it is a necessary infrastructure.
Your CRM has a record that the deal was closed eighteen months ago. That’s data. Content knowing that a deal is closed because your champion changed companies, the price had to be adjusted three times before it arrived, and that customer is now referring a few new deals a year and hates being contacted automatically. The person who operated that account knows all this. Almost no AI does, because there is almost no platform built to capture it.
This is a gap. It’s not a model gap. It’s not a data gap. Context gap. And it’s a problem that HubSpot solves with the Agentic Customer Platform. When Yamini launched the Agenttic Customer Platform earlier this year, he explained the premise behind it: a single place where all your customer data and business context resides, available to your team and your AI agents at the moment they need it.
The best infrastructure is invisible. It runs in the background, stays the same as your business changes, and doesn’t make your team duplicate itself. That’s a standard AI to keep in mind, and it’s probably not compatible.
The Hidden Cost of Content Spaces
There are costs your team pays every day that don’t appear in your AI budget. We call it the information tax: the time and repetition required for AI to produce something useful.
You define your brand voice before you ask them to write it. You attach to the account history before you ask them to search it. You define your pricing structure, your competitive landscape, your customer profile, before any meaningful activity. And the next day, he does it again. It doesn’t read your business. The real cost isn’t the hours your team loses to re-inform AI, it’s the opportunity cost: the insights the AI could come up with if it really knew your business.
Information tax is just an everyday conflict. The hard problem is the one you don’t see: what happens to context over time. Your competitive landscape is changing. Your ideal customer profile is changing. Your playbook is being updated. Your AI doesn’t know that. Not that it was forgotten. It has a chat memory. It has no connection to the business behind it.
To the GTM teams, this looks like a pretty bad AI in confidence. The project changes, your team adjusts, but the AI keeps drawing outdated content. The results are starting to be felt. The recommendations no longer match your goals.
If your AI isn’t connected to the full picture, it can’t develop the complete, powerful information it needs to create real value. It you live a a tool. It is never a reliable teammate.
Growth Teams Need Their Own Context
Not all contexts are created equal. Personal AI tools like ChatGPT create personal context: your preferences, your chat history, your communication style. Business tools like Glean form the context of an organization: your documents, wikis, and institutional information. At HubSpot, we’re building the Context for Growth: The rich, high-quality, and intuitive AI needed to drive results across marketing, sales, and customer success.
This is not a concept. We’re building a real infrastructure that will mean we’ll both capture and store this context for customers, while giving them the ability to manage it. We view the Growth Context as having five dimensions:
- Business context it’s everything you do, how you compete, and what makes you buy. Your brand positioning, your differentiation, your value base, your brand voice. This is the context that makes AI sound like your company instead of sounding like every other company. your class. Capturing it requires more than uploading a product document. It requires a system that builds that information and uses it automatically for every interaction.
- The context of the group how your people work. Your way to sell, your way to qualify, your way to climb. It’s not the version that lives in your onboarding documents, but the version that your top reps use. This is what separates an AI that follows a script from one that uses real judgment. This type of context does not exist in any CRM field. It lives in call recordings, meeting notes, and patterns that can only be seen in thousands of interactions.
- It processes context what the workflow looks like in practice. What causes a handoff. What makes a deal a priority? How your campaigns are structured and what success looks like for each one. This is what allows AI to take action, not just provide information. Building this into AI requires understanding your actual workflow, not just describing it, so the system can work on it rather than reference it.
- Total customer a cumulative history of your relationships. What each account bought, why they bought it, what their goals are, where the conflict happened, what the next logical conversation should be. This is what makes communication feel like a conversation instead of a cold call. This is a difficult category to maintain because it is constantly changing. Keeping this current automatically, across all touchpoints, is an infrastructure problem that most platforms have yet to solve.
- Total network one dimension of the Growth Context that no single company can build alone. HubSpot works with over 280,000 companies. That means we see broad trends in how teams go to market, how campaigns work, and how customers buy, on average no single company can replicate on its own. That collected intelligence becomes a layer of Growth Content available to all companies on the platform, shaping what your AI recommends before you launch a single campaign.
What the Right Questions Look Like
When testing your team’s AI, the important questions aren’t about the model. Models are increasingly being sold. The right questions are about context.
- It can hold on and do with the full picture? Not just the structured and unstructured data in your CRM, but the thinking, judgment, and institutional knowledge that often resides in people’s heads.
- Is context saved automatically? Or does your team have to keep it up-to-date manually, turning a platform investment into a maintenance burden?
- Is it specifically designed for growth? Or is it a general-purpose information layer that happens to include customer data?
- Does it add up over time? Or does it need constant replanting to stay relevant?
Answer “no” to any of these, and your AI is not working with your business, it is in your business version that is no more.
That’s a real AI race. Companies that get the Growth Context right don’t just use AI better. They excel every time they use it.
