☀️ Good Morning
Former agency owner and CMO, now building products with AI, mostly untangling big, messy data sets to find customers, keep them happy, and make the dull internal stuff run itself. Here I share my experiments, marketing insights, and the occasional misadventure in code.
Where I'm placing bets
🍝 AI on messy data
I don't want to build AI, I want to point it at big, messy data sets and real problems. Most companies sit on millions of data points they never turn into customers, retention, or saner internal workflows. That's the gap I work in.
🏗️ The new agency model
Agencies sold expensive execution. AI made execution cheap. Going AI-native isn't a nice-to-have anymore, it's the only way to ship real outcomes instead of decks. I build this way across everything I touch.
🌱 Build and back good things
I build and back a small portfolio of products around big data, sustainability and the future of work. My main one is sidestep, the tech I'm building and dogfooding across everything else. Always keen to team up with high-agency people on an interesting problem.
My modern stack
What I'm actually using day-to-day. Half marketing, half engineering.
🧰 AI pair programming
Claude Code is the engine room for every project I run now: guardrailed, context-efficient, automated workflows by default. I've built a library of custom skills it can run, everything from drafting campaign briefs to refactoring landing pages. Most of this site shipped through it.
🚂 Cloud delivery
Railway is where my projects live. Push to GitHub, it deploys. The price scales sensibly and I haven't touched a Dockerfile in months.
🎬 Creative production
Midjourney for stills, Runway for video, FFmpeg to stitch and compress. I can ship a campaign cut by lunch instead of waiting two weeks on an agency.
🧠 Memory & ops
My second brain is a folder of .md files I pipe into Claude as context. GitHub Projects is where the actual work tracks: kanban with my code already attached.
📊 Analytics
PostHog for product + marketing in one place: funnels, session replays, feature flags. The bits GA4 stopped doing well a couple of years ago.
Heavy data projects drop down into Python + Flask, with Supabase as the warehouse.
Where I keep sharp
Featured in
Want to talk?
Whatever it is, tell me below and I'll point it to the right place. I read every message myself.
Usually that's an AI build or team AI training (which I run through out-sauce.ai), growth and marketing help (I no longer take this on solo, but I work with an agency I can hand you to), or collaborating on a project: sustainability, the future of work, big data, or dragging an agency into 2026.