Work
These are the kinds of projects the studio takes on. AI handles the repetitive parts; I handle architecture, review, and delivery.
Leads come in, get summarized, tagged, and routed to the right person by a custom AI agent. Your team acts on real signal, not a raw inbox.
Customers book, pay, and reschedule themselves. Built with AI-assisted development, so you stop running the schedule out of your inbox.
One place to see jobs, invoices, and status, instead of five shared sheets nobody trusts. Built fast with AI agents and reviewed by a human.
A site that reflects what your business actually does now, with AI-assisted content and code, plus a CMS your team can update.
Example builds
These are illustrative builds, not client case studies. We’re a new studio, so rather than dress up example work as delivered projects, here is the kind of system we build, the problem it solves, and exactly which parts stay in human hands.
The problem
Enquiries arrive as unstructured email. Someone has to read every one, guess the intent, and forward it by hand. Response times slip into days and good leads go cold.
The system
An AI agent reads and tags each enquiry, routes high-intent leads to the right Slack channel, and files the rest. The team sees a structured summary instead of a raw inbox.
How we’d build it: We'd build a custom agent to parse submissions, classify intent, and post structured summaries. The routing logic, the prompts, and the edge cases stay in human hands. That part is not delegated to the model.
Discuss a similar projectThe problem
Customers book by phone or DM. Availability goes out by hand, deposits get chased, and the diary lives on paper. No-shows cost real money and nobody can see the schedule at a glance.
The system
Customers pick a slot, pay, and reschedule themselves. The owner sees everything in one place and collects payment before the appointment starts.
How we’d build it: The UI, booking logic, and payment integration would be built with AI-assisted coding. We review every API route by hand, add cancellation guardrails, and test the payment flow end to end before anything goes live.
Discuss a similar projectThe problem
Job status, invoices, and client notes live across several shared spreadsheets. Nobody trusts the numbers, and reconciling them by hand eats an evening a week.
The system
A single dashboard shows jobs, invoices, status, and revenue in real time, reading from one source of truth instead of five.
How we’d build it: We'd generate the data model, dashboard components, and chart logic with AI agents, then tighten the schema, write the auth rules by hand, and verify every calculated field against the real numbers before handover.
Discuss a similar project