Atelier
An AI-native back office for independent creatives, the venture I am building now. It starts with the pain that hurts most: getting paid on time, automatically.
Atelier gets independent creatives paid on time, without the awkward chasing. It is an AI back office that watches the tools they already use and quietly handles the admin, so their time goes to the craft, not the paperwork. The venture I am building now, with an MVP live and in validation.
Solo designers, creative freelancers and small studios are brilliant at the craft and allergic to the admin. They juggle several clients at once and run the business on a spreadsheet, a half-used accounting tool, and the dread of chasing a late invoice.
The most acute, most quantifiable pain is late payment: lost cash and lost hours. Most tools in this space are just more forms to fill in. I wanted to build the opposite: something that does the work instead of asking the creative to.
Atelier is designed to be observational, not data-entry-driven. Instead of another dashboard to maintain, it watches the tools a creative already lives in and turns the messy exhaust of running a business into clean, handled operations, surfacing a decision only when it genuinely needs the human's judgment. We are leading with the getting-paid wedge:
- It sees the invoice you emailed, tracks when it falls overdue, and drafts a firm-but-polite reminder threaded into the original conversation
- It assembles your numbers passively: outstanding, overdue and projected earnings, with expenses pulled in and categorised for you
- It collapses the path from doing the work to billing for it into a couple of clicks
Atelier is at the honest early stage: an MVP is built and we are validating it now, getting real creatives to connect real accounts and confirming the watch-and-act loop saves them the time it promises. It is a genuine zero-to-one venture, not a finished product, and I am building it with a co-founder. I scoped what to build first through structured discovery, the work behind my Compass build.
The differentiated idea is an AI split: a chat agent you can ask 'what does this client owe me?' and get a live answer and chart, plus background agents that do the chasing and the context-gathering quietly and only escalate the edge cases. That watch, act, escalate-the-exception model is the product, not a bolted-on feature. I am using AI both as the core of the product and as the way I build it, which is what lets a small team take on a problem that normally needs a much bigger one.
Taking a real, validated pain and building an AI product around it, fast, is exactly what I would do inside your business. I can go from customer evidence to a working AI tool that removes the work people hate, not just one that reports on it.
Want something like this built for your business?
I turn real problems into working products, fast, with AI. Let’s talk.
Start a conversationStart a conversation