Why we're building snill.ai
We built database tools for developers for a decade. Now we're rebuilding the same ideas for the AI age — where your business knowledge is the only specification you need.
It started with a template
My co-founder and I have been building developer tools for over ten years. First restdb.io — a database-as-a-service platform used by thousands of developers — and then codehooks.io, a serverless backend platform.
Late last year, we were using Cursor and Claude Code to build new templates for codehooks.io. One of those templates pulled together concepts we'd been refining for a decade: automatic database app creation, auto-generated REST APIs, a schema-driven UI, complex relations and lookups, computed fields, aggregations, file handling, search.
The whole thing was driven by a single JSON Schema — a DSL that defined everything. Collections, fields, relationships, dashboards, navigation, access control. The frontend and backend were completely generic. Only the schema differed between applications.
We looked at what we'd built and thought: this shouldn't just be a developer template. This could be a product for anyone.
The problem we keep seeing
Every business has operational data that doesn't fit neatly into off-the-shelf software. Client records, project tracking, equipment logs, time entries, invoices — the specifics differ, but the pattern is the same. People end up with one of three solutions:
- Spreadsheets — flexible but fragile. No relationships, no access control, no API. They break as the team grows.
- SaaS tools — expensive, opinionated, and half of the features go unused while the one thing you actually need is missing. You end up duct-taping three services together.
- Custom development — exactly what you want, but costs tens of thousands and takes months. Most small businesses can't justify it.
We've seen this firsthand with restdb.io users. Many of them weren't developers at all — they were operations managers, consultants, small business owners who used the visual builder to set up their databases and forms. They were resourceful, but the platform was still fundamentally built for developers.
What changed: the AI angle
The missing piece was always the interface. A JSON Schema is powerful, but asking a non-technical user to write one is unrealistic. We tried visual editors, drag-and-drop builders, form wizards — they all added complexity instead of removing it.
Then AI changed the equation. Today, people are used to describing what they want in plain language and getting useful results back. The conversational interface isn't a novelty anymore — it's an expectation.
So we asked ourselves: what if the specification was just a description of your business?
"I run a consulting firm. I need to track clients, projects, time entries, and invoices. Each project belongs to a client. Time entries have hours and an hourly rate. Invoices are generated from time entries."
That's it. That's the schema. The AI understands the domain, generates the data model, and you have a working app — with forms, list views, a dashboard, and a REST API — in seconds.
Why not retrofit restdb.io?
The honest answer: restdb.io's architecture is too complex to retrofit with an AI-first approach. It was built for a different era — powerful, but with assumptions baked into every layer. Adding AI as an afterthought would have meant compromises at every turn.
Snill is built from scratch on modern infrastructure. Cloudflare Workers for the edge runtime. AWS DynamoDB for the database, designed for infinite scale. R2 for file storage. Claude for the AI engine. Every architectural decision was made with the DSL-first, AI-native approach in mind.
The concepts are proven — a decade of restdb.io taught us what works. But the implementation is entirely new.
Who it's for
I want to use it myself. I need it for managing customers, invoices, timesheets, and equipment for our own business. That's always been the best test of a product — do you actually want to use it every day?
Beyond us, we think the first users will be consulting firms, startup founders, and agency owners. Not because other businesses don't need custom internal tools — they absolutely do — but because these people tend to be early adopters. They move fast, they're comfortable trying new things, and they feel the pain of expensive SaaS and spreadsheet chaos most acutely.
Snill isn't just for non-technical users, though. Every app comes with a full REST API, scoped API keys, outbound webhooks, and an OpenAPI spec. If you're a developer, you get a production-grade backend without the boilerplate. If you're not a developer, you never need to know the API exists.
What's next
We're in early access right now, refining the product with real users and real use cases. The core is solid — collections, fields, relationships, calculated fields, dashboards, access control, and the AI builder all work. We're adding features every week based on what users actually need.
If you're curious, we'd love for you to try it. Describe your business and see what comes out the other side. It might surprise you.
snill.ai is built by Codehooks AS — the team behind restdb.io and codehooks.io. Made in Norway.