About Kieran
Kieran Flanagan is SVP Marketing at HubSpot, where he has spent the last decade leading growth across the go-to-market organization. Before HubSpot, he was VP Growth at Salesforce. He started his career as a software engineer, a background that shapes directly how he thinks about AI, systems, and what marketing teams will look like in three years. He also co-hosts Marketing Against the Grain on YouTube with Kip Bodnar, and writes on Substack at kieranflanagan.io.
🎙️ Listen on YouTube, Apple Podcast, Spotify
Most marketing leaders are obsessing over AI adoption. Who’s using it, how often, which tools. But that’s the wrong conversation.
In this episode, Kieran Flanagan opened his screen and walked me through the content system he built himself in Claude Code. A SVP Marketing, live in his terminal, showing his actual workflow. Beyond that, we went deep on what AI is doing to marketing organizations structurally, operationally, and strategically. It was one of the most concrete conversations I’ve had on this topic.
Here’s what I took away.
1) The most dangerous employee in your org has high AI agency and low domain expertise. They ship fast. They ship a lot. And almost none of it is good. Kieran's view: we're in the messy middle, where AI has handed everyone superpowers, but without real craft and taste behind the output, it just generates more noise. The best people in any organization right now are the ones who combine genuine domain expertise with high AI agency.
2) AI is flattening the marketing org, but not for the reasons most people think. Marketing is structurally unlike any other go-to-market function. Sales, support, CS, and engineering all have relatively uniform roles. Marketing has always been a collection of niche specialists who largely can't do each other's jobs. AI changes that. The org gets flatter, middle management thins out and the individual contributors who remain are generalists who can run the full loop on their own, no handoffs required.
3) Measuring AI adoption in marketing is a genuinely unsolved problem. In engineering, you correlate AI usage with shipped code. In sales, it's pipeline per rep. In marketing, there's no clean equivalent. Kieran's answer : show and tells. A dedicated Slack channel where the team shares what they're building. And leaders who are AI native enough themselves to know when something is actually working and when it isn't.
4) Data quality is where AI projects live or die. The personalization workflow Kieran built at HubSpot to book meetings took three to six months to get right. Because the data wasn't clean enough. His minimum viable definition: whatever raises your AI output above the average in your specific industry. No more, no less.
5) Most B2B brands will eventually have an AI mascot. If your executives won't show their faces publicly and you don't have a creator in-house, Kieran's prediction is that synthetic AI avatars become the face of B2B brands on personality-led channels. It's already happening in B2C. A blind study he cited found that people, when they don't know it's AI, actually prefer AI-generated content because it outperforms the average human output.
Where to find Kieran: https://www.linkedin.com/in/kieranjflanagan/
His YouTube channel : https://www.youtube.com/@MATGpod
His Substack : kieranflanagan.io
Enjoy,
— Jordan
