
Whole products, pressed from scratch.
Online Olive is my studio — the design, the build, the infrastructure and the deploy. Marketplaces, internal tools, iOS apps, and the odd API an agent can pay for.
Six of them. The rest are one click away.
Client builds and studio products, all of them shipped. Hover a plate to see the real thing underneath.
Three things, held firmly.
01
The whole thing, or none of it
Product thinking, interface, backend, infrastructure and the deploy pipeline. Handing off a design to someone else's build is where most of the quality leaks out, so I don't.
02
Infrastructure that sleeps
Almost everything here runs serverless on AWS behind SST. A seasonal marketplace should cost roughly nothing in the eleven months nobody is shopping, and scale on the twelfth without a conversation.
03
AI where it earns its cost
Personalised outreach, on-device inference, semantic search — used when it changes the product, priced so the margin survives it. Not sprinkled on top for the announcement.
Smaller things, shipped anyway.
Tools, plugins, forks and one Minecraft server. Not everything needs a case study to have been worth building.
- tareInventories a Claude Code config, mines your transcripts for what you actually invoke, and moves the rest off the always-loaded path.
- Impeccable↗An expanded frontend-design skill, 17 steering commands, and a catalogue of anti-patterns to argue an LLM out of its defaults.
- CutsmithBrand-agnostic marketing content engine. Onboards a brand, pulls its real assets, invents concepts, and cuts the video locally.
- Name WizardNaming exercise as a slash command — generate a candidate pool across every naming strategy, check domains, rank into tiers.
- ThothHighlight anything on any page and it becomes a well-formed GitHub issue, matched to the repo's own issue template.
- PacerPersonal HR. Plan the week, let it pace the targets across your workdays, and never be surprised by a review again.
Got something that needs building properly?
I take on a small number of studio projects at a time — usually the ones where the interesting problem is in the middle rather than the surface.










