Episode #232: Andrew Mok — What the CMO job becomes when AI runs the mechanics

HeyGen doubled to $200M ARR in eight months, is cash-flow breakeven, and runs on about 130 people. Its CMO explains how marketing actually operates there.
For marketing leaders deciding what to keep, what to cut, and what to hand to an agent.

Andrew Mok spent 12 years at Turo building a category nobody believed in, then joined HeyGen as CMO four months ago in the middle of its fastest growth stretch. He walks through the operating details: a verticalized stack that runs inference 10x cheaper than general video models, 100 to 200 customer interviews a week piped into an internal MCP that any employee can query, and a personal agent named Mara (marketing automation and research agent) that hands him a metrics digest on his drive into the office. He's also blunt about the parts most people won't say out loud — that job replacement is real, and that it will come mostly from lean AI-native companies growing into big ones, not from big companies running layoffs. He does three to five customer interviews a week himself, and that's the metric he checks every week.

What you'll hear

• The three Cs he runs on — customers, company fundamentals, culture — and how customer voice became infrastructure instead of a slide: 100–200 interviews a week, curated by agents, queryable for pain points, churn reasons, and competitor chatter.
• How Mara actually gets used: metrics digests over voice in the car, deep dives before he reaches the office, and why documents and decks went from a multi-week exercise to 30 seconds.
• What was slowing the company down: leading with technical capability instead of the human problem, which works on AI enthusiasts and stalls with the mainstream. Plus how Turo handled its worst crisis — the New Year's terror attacks — by putting everything in the open instead of covering it.
• Jevons paradox applied to video: lower the price and friction, and quantity goes up. HeyGen's five-person internal video team produces roughly 100x more video than a company not using AI tools, and it's the biggest video team Andrew has ever run.

Chapters

• 00:00 — Cold open: AI-native companies will hire far fewer people
• 00:49 — Intro: if AI handles the mechanics, what's left of the CMO job
• 03:29 — 12 years at Turo and what long tenure teaches you about brand
• 04:34 — The New Year's terror attacks and handling crisis with transparency
• 05:44 — Turo vs. HeyGen: trusting a stranger with your car vs. a company with your likeness
• 06:52 — Who the real customer is: small business owners, skewing older, not AI enthusiasts
• 08:26 — Why he left Turo, and the introvert founder story behind HeyGen
• 09:51 — $200M ARR, doubled in eight months: what marketing changed at that velocity
• 11:09 — Breakeven economics: owning the model and running inference 10x cheaper
• 12:16 — Customers, company, culture: 100–200 interviews a week and an internal MCP
• 13:56 — Mara, the marketing agent: morning digests, campaigns, weekly team updates
• 15:27 — Which bottleneck disappeared first, and what's still hard
• 16:23 — AI and jobs: replacement comes from lean new companies, not big layoffs
• 18:21 — Jevons paradox: lower the barrier to video and demand outstrips it
• 19:50 — Hyperframes: why HTML, and why they open sourced it
• 21:56 — What surprised him inside an AI company: refusing deepfakes when allowing them would pay
• 23:17 — Lightning round
• 23:51 — The creator who kept her livelihood after long COVID
• 24:45 — Where to find Andrew
• 25:20 — Jason's top three and one action for this week

Links & resources

GuestAndrew Mok — CMO, HeyGen
Website
LinkedIn
X

MentionedHyperframes — HeyGen's open source HTML-based video rendering framework, on GitHubJulia McCoy's YouTube channel

About Frank Growth

Frank Growth is a podcast about how companies actually grow—real operators, real constraints, real decisions. Hosted by Jason Shafton.

Promotional links

• Work with Winston Francois
• Subscribe / Follow Jason