Frank Growth

Episode #241: Will Sartorius - AI didn't kill the creative job, it moved it

How to split ad creative work between AI and people so your ads stop sounding like everyone else's.
For founders and marketers running paid social who launch plenty of ads but learn little from them.

Will Sartorius, CEO of Selfmade, took over the performance creative agency after a failed acquisition left it with very little money in the bank, and rebuilt it with AI tools as GPT came out. He and Jason Shafton break down which parts of creative work AI now handles well, like tagging reviews and running time series and gap analysis on a Meta account, and which still need people: copy that makes someone feel something and angles no one has tested yet. Will walks through his "cover the logo" test, the multi-agent system Selfmade uses to write and review static ads, and why he tells lean brands to spend 50 bucks a month on a hundred AI statics before investing in storytelling.

What you’ll hear

• How to tell activity from strategy: run a time series analysis on your ad account, correlate persona, angle, and emotion to ROAS or CAC, and relaunch winning cohorts you haven't run in months.
• How Selfmade's static ad system works: it scrapes a brand's website, picks untested formats from 15 to 20 static types, runs copy through multiple copywriting agents (including one built on Daniel Kahneman's heuristics), and has 10 more agents review each image before a human signs off.
• The common mistake of copying competitors' creative, which Will says leads to "spend purgatory," and what he learned when his own top ads failed his cover-the-logo test.
• Where to start: connect the Meta API (not the Meta MCP) to Claude, tag a year of ads by persona, angle, and emotion, correlate them to ROAS and CPA, and ask what you're missing.

Chapters

• 00:00 - Intro: what's left for marketers when AI can build the strategy
• 01:28 - Will joins: taking over Selfmade after a failed acquisition
• 03:45 - What working every role at Selfmade taught him
• 05:28 - Automating review tagging, and the one skill AI can't replace
• 06:49 - Activity vs. strategy in your ad account
• 09:06 - The cover-the-logo test and "spend purgatory"
• 10:50 - Failing his own test
• 11:50 - Where humans show up: the dumbbell model and the last 5%
• 14:48 - "80% is fast, 100% is what ships": AI statics and storytelling
• 16:22 - Inside Selfmade's multi-agent static ad system
• 19:12 - New York's AI disclosure law for ads
• 20:52 - A landing page for every angle you test
• 21:32 - First fix: pull a year of Meta data into Claude
• 22:29 - Lightning round
• 24:26 - Jason's top three takeaways

Links & resources

Guest
Will Sartorius - CEO, Selfmade
Selfmade website
LinkedIn
Twitter
Newsletter

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

Episode #240: Bryan House - Tearing up the enterprise GTM playbook as a challenger brand

Elastic Path cut its website to two forms, its marketing team from 13 people to 3, and its implementation bids to a third of the price of competitors.
For B2B leaders at challenger brands whose activity metrics look busy while growth stays flat.

Bryan House is CEO of Elastic Path, a composable commerce platform competing with Shopify, BigCommerce, and Commerce Tools in a market where nearly every company already has a commerce platform. Looking at the data, he found little to no correlation between sales and marketing activity and outcomes, and not one downloaded analyst report form fill had ever turned into an opportunity. He and Jason get into cutting the top of the funnel down to two forms, putting competitors on the homepage, bringing implementation in-house to remove the incentive gap with systems integrators, and using AI and better product to move commerce implementation toward six weeks and $100,000.

What you’ll hear

• Why Elastic Path replaced BDR cold calling, nurture campaigns, and gated content with two forms on its site: get in touch or request a demo
• How marketing went from 13 people to 3, with Claude and product managers covering messaging, positioning, and sales training
• The trap of falling back on proven sales systems, like swapping BDRs for AI BDRs, when the underlying motion isn't working
• Bryan's test for contrarian moves like naming competitors on your homepage: what's the downside risk, and is it catastrophic?

Chapters

• 00:00 - Why enterprise B2B go-to-market is stuck
• 02:00 - Bryan joins: rewriting the classic enterprise GTM playbook
• 04:13 - No correlation between activity and outcomes
• 06:29 - Being ready when buyers raise their hand
• 07:51 - Commerce for weirdos
• 09:45 - Finding the right customers as a challenger brand
• 11:52 - Putting competitors on the homepage
• 13:52 - Taking implementation to six weeks and $100,000
• 16:06 - Bringing implementation in-house and the SI incentive problem
• 17:56 - Cutting marketing from 13 to 3 and letting Claude do product marketing
• 20:31 - Expansion vs. new logos
• 21:46 - Where the growth is: mid-market B2B e-commerce
• 24:17 - Lightning round
• 26:37 - Jason's top three takeaways

Links & resources

Guest
Bryan House - CEO, Elastic Path
Website
LinkedIn

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

Episode #239: Adam Spector - Why startups should automate or outsource before they hire

How to decide whether a task needs a new hire, AI, or an outside vendor.
For founders of early-stage, venture-backed companies with 5 to 75 employees who are about to add headcount.

Adam Spector is the founder and CEO of Chore, which runs the back office (HR and fractional chief of staff work) for early-stage, venture-backed tech companies with 5 to 75 employees. Chore came out of a company that raised about $10 million but couldn't raise its next round; Adam split it in two, took the services team, and set out to make it profitable. We cover why he now asks "How can we solve this with AI?" before every hire, what happened when he gave each go-to-market team member a monthly qualified-lead target after his head of go-to-market left, and why he thinks founders should outsource admin instead of building an internal team for it. One detail: Chore averages 2-3% of a customer's total monthly expenses, against about 10% at the average company.

What you’ll hear

• The question Adam asks before hiring, and the Claude project tied to Chore's CRM that now pulls the weekly sales metrics his former head of go-to-market used to compile
• How Adam reset his go-to-market team: one metric per person, a set number of qualified leads per month in each channel
• Why the team pushed back on that target, and why Adam decided it was not going to be a fit long term
• The objection Chore hears most ("that seems expensive"), and Adam's math on a roughly $1,500-a-month service versus 10 to 20 hours of a founder's time
• Growth by subtraction: why managers shouldn't be a crutch, and Adam's view that managers may be irrelevant at fast-moving companies within a year or two

Chapters/Timestamps

• 00:00 - Cold open: the go-to-market team pushes back
• 00:40 - Intro: founders have a curiosity problem, not a team problem
• 02:07 - What Chore does and who it serves
• 03:20 - Splitting the company and building a profitable services business
• 04:24 - Why referrals are Chore's main growth channel
• 05:36 - Premature hiring and asking "How can we solve this with AI?"
• 06:55 - "That seems expensive": 2-3% versus 10% on admin
• 08:33 - Resetting the go-to-market team
• 10:39 - Hiring curious builders
• 11:36 - Ollie, Chore's AI ops co-pilot
• 12:55 - Why AI doesn't have to kill services margins
• 13:38 - Growth by subtraction and managers as a crutch
• 14:43 - What managers are for in the future of work
• 15:35 - The hiring audit: automate, outsource, delegate
• 16:48 - The most expensive back-office mistake founders make
• 17:27 - Lightning round
• 18:37 - Where to find Adam
• 19:15 - Top three takeaways and one action for this week

Links & resources

Guest
Adam Spector - Founder and CEO, Chore
Chore Website
LinkedIn
X
Entrepreneurial Excellence Podcast

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

Episode #238: Ozge Erturk - Hitting #1 in the App Store on zero paid media

FloSports is profitable, covers 25+ sports, and has a million subscribers. Its CMO explains how that math works.
For subscription operators, growth leads, and anyone marketing to an audience the mainstream ignores.

Ozge Erturk is the CMO of FloSports, a profitable streaming company covering wrestling, motorsports, cheer, jiu jitsu, and D2/D3 college athletics - categories most media companies won't touch. She joined four years ago and reset the entire marketing department, expanding its scope from brand and acquisition to the full funnel, bringing all paid media in-house, and treating pricing as a growth lever. In this episode she breaks down the split between demand capture and demand creation, why she refuses to spend money manufacturing interest that doesn't exist, and what she learned running subscriptions at Amazon and Match.com Group before this. She also explains why 95% of FloSports customers sign up to watch one specific event, and why that single fact shapes everything about the pricing and retention strategy.

What you'll hear

• The demand capture vs. demand creation split, and why she's "not in the business of making you a hockey fan"
• How she reset the marketing org: full-funnel ownership, paid media brought in-house, vertical-agnostic functions
• The influencer marketing bet that didn't translate into subscribers, and her two hypotheses for why
• How to make free content and a paywall work together instead of choosing one

Chapters

• 00:05 - Cold open: the Kyle Larson test
• 00:44 - Three ways companies get niche audiences wrong
• 03:21 - Demand capture vs. demand creation
• 05:31 - What changed leaving Amazon for FloSports
• 07:46 - The marketing decisions behind profitability
• 09:29 - 95% sign up for one event: pricing and annual plans
• 10:53 - Fast Company and #1 in the App Store, with zero paid media
• 12:06 - Why the influencer hire didn't work
• 14:44 - Hiring a technical marketer, not a brand marketer
• 16:34 - What industrial engineering taught her about marketing
• 17:45 - Roku, fast channels, and the paywall strategy
• 19:25 - One playbook or one per sport?
• 20:39 - The participatory audience nobody is serving
• 21:55 - "Best kept secret in sports": feature or bug?
• 23:15 - Lightning round
• 24:55 - Jason's top three takeaways

Links & resources

Guest
Ozge Erturk - CMO, FloSports
Website
LinkedIn

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

Episode #237: Michelle Matthews - Acquisition is the easy part in health and wellness

This episode is about the gap between what marketing promises and what the product delivers, and what that gap actually costs a company.

For growth leaders, founders, and product teams building for people who show up on a bad day.

Michelle Matthews is a product and service designer with over 15 years of experience building consumer and healthcare products, at scale and from inception, including work with Brightline and Headspace. She started as a social worker, and she brings that systems view to onboarding flows, research practice, and behavior change design. Jason and Michelle cover where marketing hands off to product, why a clean divide between the two breaks the story users experience, and how to measure real change instead of engagement. She describes a user interview at Brightline where a participant walked from marketing assets through the end of onboarding and asked, are you messing with me. When a mental health company overpromises, the fallout is not just churn. It is bad reviews, lost trust, and therapists warning each other on Reddit before they apply.

What you'll hear

• The three B's from Irrational Labs: name the behavior you need, find the barriers in someone's real life, then design a benefit that counteracts each one
• How Brightline rebuilt onboarding by body storming the signup as a phone call, so the first question is how can I help you instead of what is your child's insurance card number
• Why targeting the entire mindset matrix brings in low-intent users who leave bad reviews, skip appointments, and damage lagging metrics
• The pre-research exercise: write down your assumptions, your biases, and the answer you are hoping for, then rewrite any question that leads people toward it

Chapters

• 00:00 - Cold open: are you messing with me
• 00:41 - Why acquisition is the easy part in health and wellness
• 02:50 - Social work to UX and what therapy training carries over
• 03:40 - The day someone signs up for Brightline is not a good day
• 04:55 - Selling research to a team that just wants to ship
• 05:54 - Where marketing stops and product takes over
• 07:09 - Headspace versus Calm: the Tylenol and the vitamin
• 08:31 - Engagement metrics versus real change, and the three B's
• 10:07 - Why marketers and researchers keep landing in the same rooms
• 12:08 - Where a growth-obsessed CMO gets it wrong
• 14:28 - Mindsets instead of personas, and the cost of spray and pray
• 15:42 - What overpromising costs: reviews, trust, and hiring
• 17:44 - Growing from ego versus growing in a way you can sustain
• 19:31 - Career advice for early-career designers in an AI job market
• 21:20 - Nobody is sticking around: the first thing to look at
• 22:31 - Design is not just screens
• 23:04 - Lightning round and Soft Futures

Links & resources

Guest
Michelle Matthews, product and service designer, consumer and healthcare products
LinkedIn

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

Episode #236: Elyssa Steiner - Rebuilding a 21-person marketing team in 30 days

Marketing is not a lead factory. It is a growth system, and the operating model is the ceiling on what ships.
For CMOs and marketing leaders who inherited a team built for a smaller company.

Elyssa Steiner is Chief Marketing Officer at Kahua and a B2B growth leader with more than 15 years of experience. She came from nearly nine years at Cantaloupe, through multiple acquisitions and the sale of the company, into construction project management software, and restructured a 21-person team in her first 30 days while the revenue org was verticalizing around her. She and Jason get into how she backed marketing targets into revenue targets in a workshop with the VP of sales three weeks in, why she refused to accept MQL volume as the definition of success, and how she built 14 vertical ICPs using internal knowledge and AI as a first draft. Her VP of sales called them 90 to 95 percent of the way there after 45 days, and they became custom GPTs the whole revenue org now uses.

What you'll hear

• Structuring marketing around verticals instead of channels, with solution marketers embedded in each sales pod as the vertical SME and one digital team running paid across all of them
• Backing marketing targets into revenue targets using historical conversion data, then tracking marketing sourced opportunities and closed won deals by vertical instead of MQL volume
• The mistake she names directly: scaling AI before standardizing the process, which she says she learned firsthand
• How to run the ICP exercise yourself, ship it at good enough, and let your sales SMEs throw darts at it rather than spending six months making it perfect

Chapters

• 00:00 - Why marketing teams have a throughput problem, not a strategy problem
• 02:31 - Leaving fintech for construction software, and what carries over
• 03:14 - What she found walking in the door, and why the team was hungry for clarity
• 07:12 - Keeping four vertical pods from turning into four silos
• 08:19 - Writing marketing OKRs that back into revenue, and why not MQLs
• 10:09 - Building 14 ICPs with AI in 45 days
• 12:39 - ABM from 18 accounts to 400, and the budget constraint reframe
• 14:07 - Selling AI to federal buyers when trust is the whole story
• 17:05 - Throughput across brand, product marketing, and advocacy
• 18:59 - The daily AI mandate, no layoffs, and everyone becoming a strategist
• 22:41 - Marketing to a buyer who has been burned before
• 23:38 - Lightning round

Links & resources

Guest
Elyssa Steiner, Chief Marketing Officer at Kahua
Kahua
LinkedIn

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

Episode #235: Nick Lafferty on Marketing Engineering, Category Creation, and Closing His Own Deals

He was the first marketing hire at Profound, and within weeks he was shipping production code and taking sales demos himself.

For founders making their first marketing hire and for marketers deciding what to learn next.

Nick Lafferty is the Founding Marketing Engineer at Profound, where he joined as the first marketer on a 20-person team that has since grown to 200 and raised Series A, B, and C in 10 months. He walks through what a marketing engineer actually does, what broke as the company scaled, and why he put himself in the lead round robin and gave close to a hundred demos when there weren't enough AEs to cover pipeline. He also breaks down the system he built to connect HubSpot deal data to LinkedIn impressions, the interview question he opens with, and why he is openly against token maxing. Before Profound, Nick ran growth at Loom through its acquisition by Atlassian.

What you'll hear

• Why the first marketing hire should be a marketing engineer, with a product marketer as the fast follow
• How Nick shipped a social proof quote to the live contact sales page in 15 minutes using Cursor, and the required "how did you hear about us" field that powers his attribution system
• What broke first at speed: lead routing running through a Google Sheet, deliverability issues, and then more meetings than sales reps to take them
• How Profound builds a category by publishing research to customers in Slack first, then LinkedIn, then email and webinars

Chapters

• 00:00 - Cold open: "I have contracts with real ARR tied back to my name"
• 00:45 - Frank Growth intro: three things going wrong with marketing teams
• 02:11 - Nick's background, from Loom to Profound
• 03:02 - What a marketing engineer actually does
• 03:50 - The first thing he ever shipped, and it's still live
• 04:45 - Building the LinkedIn to HubSpot attribution system
• 05:47 - The result: 20% of inbound lead forms from LinkedIn
• 06:37 - What broke first at scale
• 08:18 - Hiring the first four marketers
• 09:14 - The interview question: show me your GitHub repo
• 10:21 - What Loom taught him about horizontal products
• 12:00 - Marketing a category that didn't exist 18 months ago
• 13:23 - The research drumbeat, and giving customers early access in Slack
• 15:04 - Why Profound turns away the self-service market
• 15:45 - Profound University and the marketing engineer job board
• 16:34 - From analytics product to AI agent workbench
• 17:37 - What to look for in a first marketing hire today
• 18:50 - Anti-token maxing, and where marketing engineers come from
• 20:04 - Lightning round
• 21:30 - Jason's top three takeaways and one thing to do this week

Links & resources

Guest

Nick Lafferty, Founding Marketing Engineer, Profound
LinkedIn
X/Twitter
Profound
Profound on LinkedIn
Profound Index

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

Episode #234: Dave Steer on repositioning a brand around AI in three months

Webflow's CMO had 90 days to relaunch the website, reposition the brand, and ship an ad campaign.
For marketing leaders whose board just told them to become AI native, and who don't have a playbook for it.

Dave Steer is CMO at Webflow, leading brand, demand, and product marketing after two decades at GitLab, Cloudflare, Twitter, Facebook, and Mozilla. His mandate came from the board and leadership team, timed to Webflow Conf three months from his start date. We cover how he reorganized a 50-to-60-person marketing team around AI tools, why he ran positioning as an executive exercise with April Dunford rather than a marketing one, and where he draws the line between what a human writes and what an agent touches. He also walks through Webflow's AEO Maturity Index, a study of 2,000 US websites where the average site scored about two or three, and where the problem was site structure rather than content volume.

What you'll hear

• Why positioning is a corporate strategy activity, not a marketing activity, and who has to be in the room
• How the team moved from a relay race of specialists to treating everyone as an orchestrator and connector
• The two instincts that go wrong first: buying a pile of ChatGPT licenses, and optimizing for day one instead of day two
• Where the human/agent line sits: the human owns the argument, the agent owns the permutation

Chapters

• 00:00 - The human owns the argument, the agent owns the permutation
• 00:33 - Three ways marketing leaders are handling the AI mandate badly
• 02:24 - The three things every CMO has to be expert in now
• 03:11 - Where the three-month mandate came from
• 05:00 - What "agentic" means when the web has two audiences
• 06:14 - April Dunford and why positioning is an executive staff exercise
• 07:13 - The AEO Maturity Index: 2,000 sites, average score of two or three
• 08:40 - Rebuilt, not retooled: ending the relay race
• 12:26 - Four human roles: strategist, orchestrator, creator, connector
• 17:13 - The single biggest mistake CMOs are making right now
• 18:55 - Lightning round
• 20:49 - Top three takeaways and one thing to do this week

Links & resources

Guest

Dave Steer, Chief Marketing Officer, Webflow
Website
Webflow Conf
LinkedIn

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

Sanity stopped producing SEO content and started building pages only machines will read. Roughly 60% of last month's signups came from LLMs.

For B2B growth leaders watching organic traffic fall and trying to work out what replaces it.

Jesus Requena is CMO at Sanity, and previously led growth marketing at Figma, demand gen at Unity and Algolia, and growth at Hex. After watching client SEO traffic fall to a tenth of what it had been, he stopped SEO content production entirely and rebuilt Sanity's content for LLMs to parse rather than humans to read. He walks through the three-layer playbook: fixing contradictions across 100 to 200 pages on their own site, converting roughly 1,000 FAQ pages from user-centric to LLM-centric, building five single-topic microsites excluded from Google, then correcting third-party citations that conflicted with their own story. Sanity moved from ranking 15th or 20th in LLM visibility to top five.

• Why visibility comes from reasoning: if an LLM can reason your concepts clearly, it recommends you, which makes first-party consistency the starting point
• How they executed it — crawling their own site to find unclear terms, writing content no human is expected to read, and partnering with AirOps to fix obsolete third-party citations
• Why attribution breaks when signups arrive through a CLI or MCP without ever touching the website, and the self-reported "where did you find us" question they fell back on
• Where to start from zero, in order: audit your own site, then third-party mentions, then microsites — plus the caveat that Reddit's influence is fading and LLMs are pulling more from video

• 00:00 — Cold open: "We literally just stopped any SEO content"
• 00:38 — The AEO shift: what's going wrong and what works
• 02:12 — Meet Jesus Requena: Figma, Unity, Algolia, Hex, Sanity
• 02:45 — The common thread in marketing to technical audiences
• 04:30 — SEO traffic drops to a tenth, and the bet that followed
• 06:15 — The three layers of AEO content
• 08:22 — Citations, visibility, and going from 15th to top five
• 10:37 — Authority, and becoming the thought leader for the LLM
• 12:07 — 1,000 FAQ pages and five microsites excluded from Google
• 14:23 — Why brand investment went physical: events, swag, billboards
• 17:30 — Measuring attribution when nobody clicks
• 19:17 — Is the growth repeatable or is it timing?
• 20:41 — Where to start from zero
• 22:38 — Scaling agents, and humans handling the exception
• 23:59 — Lightning round
• 25:14 — Jason's top three takeaways

Guest
Jesus Requena — CMO, Sanity
Website
LinkedIn

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

• Work with Winston Francois
• Subscribe / Follow Jason

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