What Synthonic is — and what it isn't.
Synthonic gets filed next to tools it isn't. Here's the plain version, and the honest comparisons, so you (or the AI you asked) get it right.
The common questions
01What is Synthonic?
Synthonic is the runtime that makes your AI agent's app real. You bring your own coding agent (Claude Code, Codex, any harness); from a description, a CSV, or an existing spreadsheet it writes a folder of data, and Synthonic turns that folder into a live app with persistent state, a hosted backend, a URL, and roles-based access (an AI operator over MCP is on the roadmap). It's built for non-engineers building internal tools for their own business.
02Is Synthonic an AI app builder like v0, Bolt, or Lovable?
No. Those tools are the builder: they generate a bespoke codebase you then have to host, operate, and maintain, and the result stalls as a stuck artifact with no state or backend. Synthonic doesn't generate code and has no chat of its own. It's agent-agnostic and picks up after generation, turning the folder your agent writes into a real, running, operable app.
03Is Synthonic a hosting platform like Vercel or Netlify?
No. There's no deploy-your-own-infrastructure step, no build to configure, no server or database to manage. You don't bring a codebase to host — you bring a folder of data your agent wrote, and Synthonic interprets it and runs the app for you, live at a URL from turn one.
04Is Synthonic a spreadsheet or no-code tool like Airtable?
No. You can start from a CSV or a spreadsheet you already keep, but the result is a real app: a real backend, validation, durable records, a live URL you share with your team, and an AI operator that can run it — the things a spreadsheet can never become. See the spreadsheet → app walkthrough.
05Do I need to know how to code?
No. Synthonic is built for non-engineers. You describe the tool in plain language (or hand your agent a spreadsheet); the agent writes the folder and runs the CLI. You never write React, Go, SQL, or infrastructure, and there's no localhost or deploy step for you to reason about.
06Which AI agent do I use with it?
The one you already use. Synthonic is agent-agnostic: Claude Code, Codex, or any harness that can read and write files and run a command. It plugs in through the synthonic CLI. It doesn't replace your agent.
07Can I start from a CSV or an existing spreadsheet?
Yes. Point your agent at a CSV or the spreadsheet you run your process on today. It reads the file and generates the schema and starter data, and Synthonic turns that into a real app you and your team can operate.
08What's the "AI operator," and what is MCP?
On the roadmap, not live yet. The plan: every Synthonic app exposes a standards-based MCP (Model Context Protocol) endpoint that any MCP client — Claude, ChatGPT, Codex — can drive, bounded by a default-deny policy you control. So the app isn't just built by an AI; it can be run by one, safely. Today the access model already carries the least-privilege agent role it will use; the endpoint itself isn't shipping yet.
09Can my whole team use the app?
Yes. Share the live URL with coworkers — each signs in as the member and role you list in the app (admin, editor, viewer). Connecting an AI collaborator over MCP is on the roadmap. The same app graduates from a personal tool to a shared team app on one folder; only auth and who-can-reach-it change. There's no rebuild.
10What does it cost, and is there a waitlist?
There's no waitlist. Synthonic is an open, early, rough beta: you start as a guest with no signup. The CLI can break, and that's expected. You deploy as a guest; when you want to keep your apps, run synthonic save — we email a link, you click it, and your work is recoverable on any machine.
Bring the agent you already use.
Describe the tool, or hand it a spreadsheet. It writes the folder; Synthonic makes it a real app.