The build-vs-buy question has changed forever, at least for me

lanco.flett.cc

The build-vs-buy question for solo freelancers and many businesses alike used to be easy: buy something and live with the compromises. Either you get bloated enterprise software with seven hundred features you'll never touch, or you get basic tools that fall apart the moment you have multiple clients, multiple currencies, or any complexity at all. Building wasn't really an option unless you fancied spending six months not doing actual work.

Turns out you can build now exactly what you need in a couple days, and it costs about as much as three months of a SaaS subscription you'd cancel anyway.

AI as development and product strategy

Lanco is an experiment in how to build software with AI, and if that changes how I decide which tools to use when I'm not building.

The codebase is almost entirely AI-generated, with me steering, reviewing, and refining. Community feature requests come in through the roadmap, get translated into prompts, and turn into pull requests that I review before merging.

It's a test of how far this can go. Can a single person maintain a real product this way? What's the right balance between AI generation and human oversight? I guess I'll find out.

Wait, what is lanco?

Oh yeah... Lanco is a freelance management tool:

Projects and clients organised in one place, with custom day rates and clear structure
Time logs and invoices that flow naturally from tracked hours to invoices in a single click
Expense tracking by category and client, so you know exactly where the money goes
MTD reporting with CSV exports ready for HMRC submission
AI-powered documentation to generate and refine proposals, contracts etc

The interface is mobile-first and deliberately minimal. Multi-currency support, historical exchange rates, and inline reporting are all there, just not screaming for attention.

As an aside: AI can build, but it can't delight

Something I've discovered for myself during this process is that AI is excellent at producing functional code. It can scaffold features, wire up APIs, and handle the boring plumbing faster than I ever could. But it has no taste.

Left to its own devices, AI produces interfaces that work but don't feel right. The animations are either missing or generic. The micro-interactions that make software feel polished, the subtle hover states, the satisfying button feedback, the smooth transitions, none of that comes naturally to AI. It builds to spec, not to delight.

This is where human design oversight becomes essential. I review every AI-generated component for feel, not just function. Does the invoice slide in or just appear? Does the delete confirmation feel appropriately weighty? Does the whole flow feel like one coherent product or a collection of features?

The AI handles velocity. The design eye handles quality. Both are necessary; for now anyway.

Opening it up

Lanco started as something I made for myself. Now it's free to use and growing through community contributions.

Give it a go at lanco.flett.cc.