How HubSpot's CEO (Yamini Rangan) Thinks About AI & GTM

In January 2023, Yamini Rangan scrapped HubSpot's entire product roadmap. Not parts of it. All of it.

That decision, made at a public company with almost 300,000 customers, is one of the most instructive examples of conviction under pressure I've come across in recent memory.

Why This Conversation Matters

I've been fortunate to sit across from a lot of operators on Growth Unlocked. Revenue leaders, founders, GTM practitioners who've built serious businesses. But getting Yamini in the chair felt different.

She's been CEO of HubSpot for five years. She navigated a pandemic, a market correction, and one of the most significant platform shifts in software history. And she's still standing, still building, still figuring it out in real time.

What struck me most about our conversation wasn't the scale of what she's achieved. It was how human and unpolished the journey sounds when she describes it. No clean narrative. No tidy playbook. Just conviction, iteration, and a relentless focus on one thing.

If you're a founder or revenue leader trying to work out how to respond to AI right now, this is the conversation you need.

One North Star, Everything Else Filters Through It

The first thing Yamini said when I asked what the last five years taught her was this: focus on one North Star and never let go of it.

For HubSpot, that North Star is solving for the customer. Full stop.

It sounds simple. It is simple. But she was clear that the simplicity is exactly what makes it powerful. When a new model drops, when a new tool emerges, when a competitor makes a move, the filter doesn't change: what does this allow us to deliver for our customer?

That lens cuts through noise fast. And right now, there's a lot of noise.

I've felt it myself. New AI tools are coming out at a pace that's hard to keep up with, and the temptation to chase all of them is real. Having a clear filter for what actually matters to your business is what stops you from scattering your energy across things that don't move the needle.

The takeaway: If your team is struggling to prioritise AI investments, the answer isn't a better framework. It's a clearer North Star. Get that right and the decisions downstream become significantly easier.

Incremental or Platform Shift? The Question That Changes Everything

When ChatGPT launched in November 2022, Yamini was one of the 800 million people using it. Every day. By the time HubSpot got to their Q1 2023 board meeting, they had a very clear answer to one question: is this technology incremental, or is this a platform shift?

They concluded it was a platform shift. And that single answer made everything else obvious.

If it's incremental, you keep the roadmap and add some features. If it's a platform shift, the roadmap is irrelevant. The entire trajectory of the company has to change.

This is the question most business leaders are still sitting on the fence about, and the cost of that ambiguity is high. You can't half-commit to a platform shift. You either recognise what it is and respond accordingly, or you get caught optimising for a future that no longer exists.

For HubSpot, that meant launching their first AI features in March 2023. Two months after the board meeting. That's a fast cycle for a company of that size.

The takeaway: Ask yourself honestly, is AI a feature you're adding, or is it changing the fundamental nature of how you deliver value? Those are different problems with very different responses.

The Agentic Shift: Software That Does Work, Not Just Helps You Work

One of the clearest frameworks Yamini shared is the distinction between software that helps you get work done, and software that does work for you.

The previous decade of software, including most of what we think of as CRM, was designed for humans to use. You logged into it, you entered data, you ran reports, you took action. The software made the human more productive.

What's changing now is that the software is starting to take action on your behalf. Agents that research accounts, prioritise outreach, draft emails in your voice, update the CRM after a call without you touching it. That's not an improvement in productivity. That's a different category of tool entirely.

As Yamini put it, the shift is from software helping you get work done to software doing work for you. And that has enormous implications for how go-to-market teams operate.

I told her about my experience running Claude through a terminal connected to HubSpot via MCP. In nine hours, I had done deep vertical research, built a long-form content piece, identified an industry association with access to our ICP, written personalised outreach to decision makers, and built an interactive lead generation tool for our website. Work that would normally stretch across a week.

That's not productivity. That's leverage. And it's available right now to anyone willing to set it up.

The takeaway: Stop thinking about AI as a feature in your stack. Start thinking about it as a workforce you're configuring. What would you delegate to a tireless, well-briefed team member who works 24 hours a day?

The Gap Between Output and Outcomes

Here's the thing that most companies are missing, and Yamini put her finger on it precisely.

Lots of teams are generating AI output. Very few are driving AI outcomes.

She described talking to customers who had adopted AI tools, generated more content, more emails, more reports, and ended up with more volume but not more growth. The AI just added noise.

The gap between output and outcomes comes down to context.

Context about who you are within the business, what you're trying to achieve, who your ideal customer is, how your brand communicates, what industry-specific language matters. A generic AI prompt doesn't know any of that. It produces generic output.

This is what Yamini described as the core architectural shift HubSpot is making with what they're calling the agentic customer platform. Structured data, unstructured data from calls and emails and meetings, all indexed and made available so that the agents operating on your behalf actually understand the context of your business.

An agent without context is like a new hire who's never been through onboarding. Technically capable, but producing work that misses the mark.

The takeaway: Before you scale AI across your GTM team, audit your context. Is your ICP documented clearly? Is your brand voice codified? Are your customer conversations captured in a searchable format? The quality of your AI outputs is a direct function of the context you've built.

Speed With Direction: The Only Kind That Counts

Yamini said something in our conversation that I keep coming back to.

Speed without direction is chaos. Speed with clarity of direction is a force multiplier.

She described HubSpot going through a phase of telling everyone to run, letting a thousand flowers bloom, people experimenting across every Slack channel, building things, sharing what they made. And then stepping back and realising they were in the same place. Energised, but not progressed.

They had to pull people back, name four clear priorities, and point the entire organisation in one direction.

This maps to what I see with a lot of growth-stage companies right now. There's enthusiasm for AI. There are experiments running. But there's no coherent direction that makes the experiments compound. Individual productivity goes up, but institutional productivity stays flat.

Moving fast matters. But direction is what turns velocity into results.

The takeaway: If your team is experimenting with AI in a hundred different directions, that's phase one. Phase two is picking four things, going deep on them, and making those compound across the whole organisation.

Working Backwards From the Future

I asked Yamini about imposter syndrome and self-doubt, because she's openly talked about it in the past and I think it's one of the more underrated leadership topics.

Her answer was practical rather than philosophical.

Instead of looking at her past capabilities and asking what she's capable of in the present, she looks at the future she wants to create and works backwards. What does HubSpot need to look like? How does she want to show up as a leader? What actions today close the gap between where she is and where she wants to be?

It's a reframe I've found genuinely useful. Self-doubt tends to be anchored in your past, in what you haven't done before, what you've got wrong, what you don't know. Future orientation cuts through that. The question isn't whether you've done this before. The question is what needs to be true for the outcome you want to exist.

For founders navigating an AI-driven market with no clear playbook, this is a useful mental model. You won't find the answer in your history. You have to build towards the future you're trying to create.

The takeaway: When uncertainty creates paralysis, switch the lens. Don't ask what your experience says you're capable of. Ask what needs to be true for the future you want to build, then figure out what you need to do today.

Inaction Is the Biggest Risk

I want to finish on the piece of advice Yamini gave for founders and revenue leaders who are watching things shift and waiting for clarity before they move.

She was direct about it: don't wait for the right answer. Learn your way into it.

The analogy she used was one-door versus two-door decisions. Two-door decisions, the ones you can reverse if they're wrong, should be made fast. Go, test, learn, adjust. You don't need a full analysis for a reversible call. Inaction on a two-door decision is just wasted time.

One-door decisions deserve more thought. But even those require urgency. Because inaction, she said, is terminal right now.

That framing resonated with me. A lot of the AI decisions founders are agonising over are actually two-door. You can try an agent workflow, see if it adds value, and change course if it doesn't. The cost of experimentation is low. The cost of sitting still and watching competitors build advantage is high.

The companies that will look back on this period with pride won't be the ones who waited for certainty. They'll be the ones who got into the arena early, learned faster than everyone else, and used that learning to compound.

The takeaway: Most AI decisions are reversible. Stop treating them like they're not. Move, learn, adjust. The only genuinely risky move right now is standing still.

Closing Reflection

I started Growth Unlocked because I wanted to have the conversations that actually matter for founders and revenue leaders. Not surface-level tactics. Not self-congratulatory success stories. Real thinking from people who are in the middle of it.

Yamini is one of those people. Five years into one of the most demanding CEO roles in SaaS, navigating a platform shift without a clear playbook, making bets with incomplete information, and being honest about what she doesn't know.

The clearest thread across our entire conversation was this: the leaders who will navigate this period well aren't the ones with the most certainty. They're the ones who are most willing to learn.

That's the Growth Unlocked philosophy in a sentence. Clarity comes from doing, not waiting. Conviction comes from having a clear North Star, not from having all the answers.

Go build. Learn fast. Stay grounded on what matters.

Key Takeaways

1. One North Star cuts through everything. When AI tools are multiplying daily, a single clear customer-centric filter makes decisions faster and better.
2. Ask the right question early. Is this incremental or a platform shift? The answer determines whether you iterate on your roadmap or scrap it.
3. Software has changed roles. The new question isn't how do you make your team more productive. It's what can you delegate to agents operating on your behalf.
4. Context is what separates output from outcomes. Generic AI produces generic results. The companies winning are the ones who've invested in the context that makes AI outputs actually useful.
5. Speed needs direction. Letting a thousand flowers bloom is phase one. The next stage is pointing the whole organisation in one clear direction and compounding from there.
6. Work backwards from the future. Self-doubt is anchored in the past. Future orientation breaks the paralysis.
7. Inaction is the biggest risk right now. Most AI decisions are reversible. Treat them that way. Move, learn, adjust. The cost of standing still is high.