How does a growth operator who built and managed large teams at Twitch and Discord work today as a solo fractional CMO?

In this episode of Intelligent Artifice, Shamanth Rao sits down with Justin Gerrard, fractional CMO and former growth lead at Twitch and Discord, to talk about how he uses Claude day to day to plan his week, build GTM plans, and deliver work for multiple startups at once. Justin shares where AI genuinely saves time, where human judgment is still the difference maker, and why letting AI run on autopilot is a trap most teams fall into.

If you work in growth, GTM, or performance marketing, this is a practical and honest look at what working with AI actually looks like in 2026.

Chapters:

00:00 Intro: Justin's background at Twitch and Discord
01:21 Building performance marketing at Twitch
03:22 AI Speed: Weeks of work vs. a few hours
05:54 AI as an Operating System
06:33 Justin's Monday Morning AI routine
08:05 The 60/70 Rule for AI productivity
09:12 Why 100% AI work is a hiring red flag
10:11 Case Study: A localized GTM plan
14:52 Creative Strategy: Automated UGC at scale
16:30 The Trap of Autonomy in AI content
18:13 Brand Safety: The 18 Curses story
19:15 The Librarian vs. The Author
21:35 Human Creators vs. AI Avatars A/B Test
24:55 Where to find Justin online

Key things we cover in this episode:

✅ How Justin structures his Monday morning using Claude Desktop and MCP connectors
✅ Why he caps AI contribution at 60 to 70% on every GTM plan
✅ Why AI is a fast librarian but not an author
✅ How he A/B tested AI creators vs human creators in a paid ad campaign and what the results showed
✅ Why fully automating content output does not drive real results
✅ What experienced operators still bring to the table that AI cannot replicate

Web funnels are like teenage hex.

Everyone claims they are crushing it. Almost nobody really has a clue.

Elise Zareie spent last year actually building them.

She has been in UA since 2019. Last year she became a part-time product manager just to ship funnels herself. That meant learning Figma, coordinating designers, front-end developers, back-end developers, and analytics teams. Running QA. Shipping it. Then using AI to test faster than she ever could before.

In this episode she talks through what the process actually involves, the three levers that move the needle in any funnel, the one benchmark she watches obsessively on landing pages, how she uses Claude to generate full funnel copy from screenshots of top-performing creatives, and why AI visuals are making consumers more suspicious, not less.

Key topics

• Why web funnels are harder to build than most people think and what the process actually looks like
• How to identify the dominant funnel in a vertical before building anything
• The 40% page-one to page-two benchmark and what to do when you fall below it
• How to match landing page copy to ad creative using UTM tags
• How Elise uses Claude to generate funnel copy and assessment questions from top-performing creatives
• How AI helped her launch a male-specific funnel in two weeks for an app with 80% female users
• Why AI-generated visuals are creating consumer suspicion and what to do about it

In this episode, I'm joined by David Barnard — indie app developer, RevenueCat Developer Advocate, and host of the Sub Club podcast — to talk about what happens when someone with 18 years of App Store experience picks up vibe coding for the first time.

David has launched 30+ apps, sold four, and seen every major shift in the App Store since day one. Now he's building a new app entirely on his own — no contractors, no coding partners — just Claude Code, Codex, and decades of product intuition.

This conversation goes deep on what's actually changed, what still requires human judgment, and why distribution matters more than ever in a world where anyone can ship an app.

What we cover in this episode:

- Why vibe coding finally clicked for David — and what changed in the last 12–18 months
- How 18 years of App Store experience shapes what you can build with AI
- Claude Code vs. Codex — how David uses both and plays them off each other
- Why product intuition is still the hardest thing to replicate with AI
- The real cost of building an app in 2025 vs. 2020
- Why distribution is the new moat — and how David is thinking about it
- RevenueCat's Rico: talking to your business, not just your data
- Why "if you build it, they will come" has never been true on the App Store

Whether you're an indie developer, founder, or operator, this episode will reshape how you think about building and shipping in the AI era.

Content Update: Rico is no longer just in beta! You can now access these insights directly in your RevenueCat dashboard.

Learn more:

— Podcast hub
🌐 https://mobileuseracquisitionshow.com/ — Podcast hub
🌐 revenuecat.com — Subscription infrastructure for mobile apps
🎙️ https://www.youtube.com/@SubClubHQ - App business insights from the people building them