I engineer attention for a living.

Scripting the Brewed Podcast

Brewed is Appbrew's podcast for DTC founders and operators. Guests include the leadership behind brands like MiaMily, SUKOSHI, Svaha USA, Felina, and Kut from the Kloth. My job: make a 60-minute conversation feel effortless while secretly being engineered to the minute.

Every episode gets a full research pass before a single question is written. I scrape and read the guest's entire LinkedIn post history to learn their actual voice, pull their conference talks, and mine the brand's app analytics for numbers worth asking about. Then I build a 20 to 21 question script organized into chapters, each chapter a distinct act of the conversation.

The scripts aren't question lists. They're designed so the guest's best stories are unavoidable, the clips cut themselves, and the host always has a bench of follow-ups ready. I also ideated the podcast's landing page for distribution, live at appbrew.com/podcast, so every episode has a home that converts listeners into subscribers.

Read the full case study →

A LinkedIn system built on frameworks

I turned Appbrew's LinkedIn output from ad-hoc posting into a documented system. Every carousel and post is built on a named framework: AIDA, PAS, BAB, or 4Ps, chosen deliberately based on whether the piece is educational, thought leadership, or story-driven.

Then I went further and encoded the whole system into two custom AI writing skills with reference libraries for hooks, psychology, CTAs, and best practices, so every piece of content passes the same quality gates no matter who runs it. That's the difference between a writer and a content operator: I build the machine, not just the post.

Every output runs through an EEAT self-check before publishing: does it show real experience, teach something actionable, make a direct claim, and sound honest? Weak on any pillar, it gets rewritten.

A community built from zero: 174K views, 4.3K members

I run a Shopify growth community on Reddit for Appbrew. Over the past 12 months it has generated 174K views and grown to 4.3K members, with 4.2K joining and only 73 ever leaving. Churn that low means the positioning pulls exactly the right people.

The sharpest sprint was March 2026: members went 208 to 555 in 30 days and views jumped from 8.2K to 19.3K, with one thread spiking 4.5K views in six days. I track it all weekly, down to daily uniques (now averaging 3.2K) and weekly visitors (3.7K).

The healthiest signal isn't the member count. It's that 8.7K comments were published against 2.2K posts, a 4:1 ratio. People aren't just joining. They're arguing, answering, and coming back. And moderation stays strict: 988 low-value posts removed to protect quality.

Read the full case study →

Video ads built from a tested hook library

I write 30-second video ad scripts for Meta Reels and LinkedIn using a library of 18 tested hook types: contrarian, conversational, "why did no one tell me", trigger-word, demographic, and more. Every hook is built in four layers: text overlay, sound, visual, and vibe, because most ads are watched on mute and the visual hook usually matters more than the words.

Each script ships with 4 rotating hook variants to A/B test, scene-by-scene breakdowns timed to the second, silent-scroll captions, and platform-specific swaps (a "premium customer" data clip for LinkedIn, a warm buyer-voice clip for Meta).

My signature move: mining our own podcast transcripts for founder quotes and mapping them into the ads, so real customer voices do the selling instead of a voiceover making claims.