What a decade of restdb.io taught us
restdb.io has always been a no-code tool. You log in, click "Create New", and start defining collections, fields, forms, and pages in a visual admin UI. No code required. The REST API, the MongoDB-style query syntax, and the JavaScript codehooks are there if you want to extend something, but they are optional. Most of what people build with restdb.io is built entirely in the browser.
Over the years, restdb.io has been used by thousands of people who are not developers: operations managers, consultants, agency owners, small business founders. People who outgrew spreadsheets but couldn't justify a custom build. We've watched them build all kinds of things, and we are still slightly amazed by some of it.
The common thread is that they all had to model their domain in the system. Collections, fields, relations, validation rules. restdb.io takes care of the database, the API, and the admin UI for you, but the design work is still on you. That's the deal with database tools, no-code or not. For a long time, that was as good as it got.
What changed
AI has changed software development completely in the last two years. Not just in some areas. Everywhere. How code gets written (we use Cursor and Claude Code daily, and we are not going back). How features get specified. Who can build software at all. What counts as a "specification" in the first place.
For most of our careers, building software has meant translating human intent into a precise, structured form: code, a schema, a workflow, or a stack of validation rules clicked into an admin UI. Either way, you have to think in the system's terms, not your own. That translation cost shaped what got built and what didn't. Entire categories of software (small, specific, business-fit) just never made the cut.
That cost is collapsing fast. Describing what you want in plain language and getting a useful result back is no longer a novelty, it's an expectation. The trade-offs that have shaped our industry for decades are changing with it.
restdb.io is a tool from the era before that, and that's fine. It does what it does well. But the next generation of internal-software tooling is going to look different: you describe your business, and an AI builds and runs the system for you. We started seeing that shape last year while building new templates inside codehooks.io (our serverless backend platform). One template pulled a decade of ideas together into a single JSON Schema DSL: collections, fields, relationships, dashboards, navigation, access control, all driven from one specification. The frontend and backend were generic. We looked at it and realized: if the specification could be a description instead of a schema, the modeling step was gone.
Introducing snill.ai, the next generation
snill.ai is what we built next. It is the next generation of what we've been working on for a decade, and much more ambitious than restdb.io ever was.
You describe your business in plain English (your clients, your projects, what you bill for, how you track work), and snill builds a complete working system for you. Not a database with forms on top. A live internal application, with:
- A relational data model (collections, fields, lookups, computed fields), generated from your description
- Dashboards tailored to your business, not picked from a template
- An AI agent that already knows your data, your model, and your business context, because it built them. Ask it questions in plain language, get real answers
- Live evolution: tell snill "add a status field to projects" or "give my ops lead a dashboard of overdue tasks", and the system updates itself in place
- Version history and self-service backups: one-click rollback, daily automatic backups kept for 30 days, manual snapshots before risky updates
- Team access with custom roles and permissions at the collection and field level
- A full REST API with an OpenAPI spec, scoped keys, and HMAC-signed webhooks (optional, if you want them)
If you've used restdb.io, you'll recognize parts of it. The same concepts we've refined for a decade (collections, relations, computed fields, APIs, hooks) are still in there. But snill.ai goes well beyond what restdb.io ever did. The AI is a first-class part of the product, the dashboards are generated for your work, and the system grows with you in conversation. It also ships with a few things we always wanted in restdb.io but never got around to, like self-service backups and a proper team access system with field-level permissions.
snill.ai is a fresh build on a fresh stack (Cloudflare Workers, AWS DynamoDB, R2, and Claude as the AI engine), not a retrofit of restdb.io. We chose not to bolt this onto restdb.io's older architecture, because it would have meant compromises everywhere.
Not vibe coding
One thing we want to be clear about. snill.ai is not vibe coding. Our goal is the opposite: to make building internal business software completely safe.
We are not generating arbitrary application code that you have to read, debug, and maintain. snill builds a structured application (a real data model, database, UI, and API), and the snill platform runs it. Every change is versioned, so you can roll back any update with one click. There is no "the AI rewrote everything overnight and now nothing works" failure mode.
For custom logic and integrations beyond what snill itself does, you don't drop into a code editor. You wire snill to the rest of your stack through webhooks and API calls. For custom serverless code, send events to codehooks.io. For workflow automation, connect to n8n, Zapier, or Make. snill is the structured core. The rest of your stack plugs into it.
What this means for restdb.io users
restdb.io is still running. Your data, apps, and APIs are not going anywhere, and we are keeping the lights on.
But the place we're investing now (and the place we'll point anyone who's ever wished they could skip the schema step or hand a system to a non-technical teammate) is snill.ai.
The full origin story is on the snill.ai blog: Why we're building snill.ai. It's worth a read if you want to understand where this is going.
