Kieran Flanagan on Why Growth Teams Will Disappear at HubSpot

The AI Innovation Pod Will Replace Growth Teams

Growth teams exist to make go-to-market touchless. Onboarding, activation, upgrade flows—all the ways to extract humans from the buying process. But Kieran Flanagan thinks that model is dying.

"What I think AI does is extrapolate all that away to just like, how do we make our go-to-market much more AI-centric?" he says. The future isn't a growth team optimizing product tours while sales optimizes their pitch deck. It's one team—an AI innovation pod—deploying agents across the entire go-to-market, whether that's self-serve onboarding or customer support or sales.

Who owns that experience, right? Like what's the handover? What part does growth own? What part does customer success own? What part does sales own? To me, an AI innovation pod should be the pod that's trying to integrate that experience across your go-to-market.

— Kieran Flanagan

Picture a multimodal agent that meets you on a website, guides you through onboarding via text or audio, sees your screen, and stays with you through the upgrade and into customer success. The old boundaries—growth vs. sales vs. support—collapse. HubSpot hasn't disbanded its growth team yet, but Flanagan sees the writing on the wall. If you're starting fresh today, skip the functional silos. Build one team that can think across touchless and human-led motions.

How HubSpot 5x'd Email Conversion With Style Prompts

Personalization works. Not the "Hi {{FirstName}}, I saw you work at {{Company}}" kind. The kind where AI extrapolates your conversational style and rewrites the email to match it.

HubSpot broke a single email into separate prompts: one for subject line, one for intro, one for body, one for close. Each prompt pulls from structured and unstructured data—sales transcripts, chat logs, behavioral signals. Then they run the email through a style generator that learns what converts you specifically, not your segment.

AI is able to extrapolate styles. So it's able to say, hey, like there's a certain style that makes up this email and I can replicate that style for you.

— Kieran Flanagan

The result? Triple-digit conversion lifts. Flanagan admits he wouldn't have believed it 12 months ago. But the key is data. The better your data layer—internal and external, structured and unstructured—the better the output. Segment-level marketing (job title, seniority, industry) is dead. AI lets you market to 300 people with 300 styles. The constraint was always headcount. Now it's creativity.

Outbound Email Now Takes 3–5x More Volume to Book the Same Meetings

AI made everyone pretty good at email. That's the problem. Flanagan has seen the data: it takes 3 to 5 times more emails today to book the same number of meetings as a year ago. Supply went infinite. Conversion tanked.

The winners won't be the ones sending more AI-generated emails. They'll be the ones using data creatively to stand out in the noise. Most people take what AI gives them and ship it. The 20% that matters—the style, the hook, the insight—still requires exceptional humans.

There's going to be in every single discipline in terms of go-to-market, 20% that really matters. And it's like, can you use AI to do most of the other 80% and have the exceptional humans figure out how to make the 20% much better, much different, much more unique.

— Kieran Flanagan

But even being great might not be enough. When Lovable builds 10% of all websites in a single month and distribution channels (Google, paid ads) are shrinking, you have exponential supply growth colliding with collapsing distribution. That's the real crisis.

The Content Collapse: Why Informational SEO Is Over

Cloudflare's CEO shared a stat that should terrify content teams: Google used to scrape 2 pages and send 1 visit back. Six months ago, it was 6 pages per visit. Today it's 18 pages scraped for every 1 visit. ChatGPT? 1,500 pages scraped per visit.

AI Overviews now appear on 35% of Google searches. Google announced AI mode will become the default search experience. The incentive to publish content is evaporating. Why create if no one clicks through?

Flanagan splits search into two buckets: informational (teaching content) and transactional (bottom-of-funnel, high-intent). Informational is dead. People will get answers from AI assistants without ever visiting your site. Transactional still has value—people searching for a product to solve a problem will still click. But the days of building a SaaS company on blog traffic are over.

HubSpot, a company built on organic content, is adapting. They're not abandoning SEO, but they're realistic. The relationship between publishers and AI assistants needs to be rebuilt. Until then, focus on transactional intent and figure out how to show up inside AI assistants, not just Google.

AI Tools Are Eating Labor Budgets, Not Software Budgets

Flanagan predicts we'll start pricing AI agents against human labor costs, not software seats. HubSpot redeployed customer support teams to white-glove onboarding for premier accounts after AI handled tier-one support. BDRs can book the same meetings with fewer people—or book more meetings with the same team.

The biggest hype? Autonomous multitask agents. They're inconsistent and unreliable. The models aren't there yet. The biggest wins? Personalization, customer support, and selling through chat. HubSpot's AI chat agent now books meetings and does sales, not just deflects support tickets.

But here's the truth no one wants to say out loud: teams will be smaller. AI gives you time back. How you redeploy it—better unit economics or higher-value work—is a choice. Either way, the headcount chart is going down.