What is a Solo+AI studio?
A solo+AI studio is one experienced product person using AI tools as a workforce — delivering the output of a traditional multi-person studio, from strategy and UX through deployed code, with zero handoffs and a single point of accountability.
The term describes a working model that became viable around 2024–2025, when AI coding agents (Claude Code, Cursor) and AI design tooling crossed the threshold where one senior person could produce ship-grade volume across disciplines. Not a freelancer with ChatGPT. Not an agency that "uses AI." A structurally different unit: senior judgment at every decision, AI at every keyboard.
What makes it a distinct model
- Zero handoffs. The person who interviews users is the person who designs the screens is the person who writes the frontend. Nothing is lost in translation because there is no translation.
- Outcome scope, individual structure. A freelancer sells hours inside one discipline. A solo+AI studio sells the full loop — research → design → build → deploy — the scope that used to define an agency.
- AI as workforce, not feature. The AI does the volume work that justified junior headcount: component variants, boilerplate, test coverage, refactors. The human does what AI can't: deciding what's worth building and what good looks like.
- One name on the work. No account manager between the client and the maker. Accountability doesn't diffuse across a team because there is no team.
The economics, honestly
The traditional agency price covers layers a client never sees: account management, project management, internal reviews, junior staffing learning on your budget. The solo+AI model removes those layers rather than discounting them.
| Freelancer | Solo+AI studio | Agency | |
|---|---|---|---|
| Scope | One discipline (design or code) | Full loop: research → design → build → deploy | Full loop |
| Handoffs | To someone else's developer | None | Several internal + one to client's team |
| Speed to MVP | Partial output only | 2–8 weeks | 3–6 months typical |
| Cost shape | Hourly, open-ended | Fixed scope, fixed price | Retainer or large fixed bid |
| Accountability | One person, narrow scope | One person, full scope | Diffused across team |
| Best for | Filling a defined gap in a team | MVPs, first versions, AI products needing speed | Multi-team programs, guaranteed parallel capacity |
When NOT to use a solo+AI studio
Honesty is the point of this page. The model has real limits:
- Very large programs. If the work genuinely needs five workstreams in parallel for a year, you need an agency or an in-house team.
- Bus factor. One person is one person. Mitigated by documentation and clean handover, but not eliminated.
- Procurement requirements. Some enterprises can only contract firms with bench depth, certifications, and an org chart.
Where the model came from
This page is maintained by Aleksey Stepikin, who runs Stepikin Studio on exactly this model after three years as Head of Design at Diagnocat (AI dental platform, 1000+ clinics). Working proof of the model: Vigilo — a global risk monitor with 44 live sources across 198 countries, built solo; and EngiBoard — a 4-platform desktop SaaS shipped in 2 weeks (build diary).
For the interface patterns this work produces, see LLM UX Patterns. For pricing and process, see the FAQ.
Building an AI product and weighing agency vs freelancer vs this?
Book a free 30-min call