AI & Workflows.

Most AI rollouts fail because they speed up a legacy process instead of changing it. Notes on what actually shifts when you redesign roles alongside the work, from a five-person team where everyone went full stack, and why the same move is so much harder, and so much more necessary, at enterprise scale.

I used AI to build an automated daily puzzle and a marketing machine around it. New puzzle at midnight, screenshot to five social networks at 08:00, reminder email at 06:00, stats dashboard refreshing itself. I went on holiday for a week. It kept shipping.

Specialists get compressed into agents. Generalists with taste, deployment scars, and a domain get compounded by them. Field notes on the role modern companies are quietly starting to need.

Two AI content pipelines running at Sourceful. A Veo 3 video-ad workflow that replaced a production agency, and a six-agent OpenClaw system that researches, writes, publishes, and promotes blog posts end to end. Field notes, including the morning it trolled a supplier.

A few hours in Claude Design, a rebrand to roll out, and a reasonably honest review.

Building a marketing site that serves four different audiences, welcomes AI crawlers, and prioritises accessibility from day one. Three weeks, two people, no agency.

34 hours of actual work, 50+ production-ready components, and a story about where all that computational power actually comes from.