Your work stays
under your control.
Your code changes only when you ask. Roles are decided on the server, sharing is set per file, and AI is a company switch.
Most of the answer
is the shape of it.
- Nothing changes until you ask.
Connecting lets Lindi read your project and put its screens on the canvas. Editing the design does not change the code: the code changes when a developer accepts a design you push, or when you ask Lindi Code to build.
- Agents act as you, with your access.
An agent connected over MCP reads the files you can open, and can edit the ones you may edit — in your open tab, or in the background unless you turn that off in Settings → Preferences.
- A token you issued, yours to revoke.
Every connection uses a token you create in Dev Mode. It works as you, on every file you can open, so keep it like a password. Revoke it and it stops working the same moment.
Nothing ships
without your team.
What Lindi writes lands in your project as ordinary changes to its files, so your usual workflow — diff, review, commit — covers it. Nothing reaches your users until your team ships it.
- Code changes only when someone asks
- Your usual review covers every change
- Your team decides what ships
Who may change things,
and who only looks.
- One role per person, decided by the server.
Everyone holds exactly one role, and what they may do comes only from that role — checked on the server every time, never taken on trust from the browser.
- Owner, editor, viewer.
Sharing is set per file — a person, your company, anyone with the link, or nobody. A viewer can view, inspect and comment, but not edit the design.
- Libraries your company approves.
Choose the libraries everyone in the company works from, so the design system in use is the one you chose.
AI, on your terms.
The questions every organisation asks about AI, answered in the product rather than in a policy document.
- An owner can switch it off.
AI is a setting on your company, not a personal preference. An owner can turn it off for everyone, and every change to the setting is recorded.
- No model keys on your machine.
When Lindi Code writes into a project on your machine, its model calls go through Lindi — no provider key is stored on your device.
- Models named, not hidden.
Lindi Code runs on Anthropic’s Claude, and you can see and choose the model. The image tools run on Google’s Gemini. OpenAI turns voice input into text and translates canvas text.
Our cloud,
or your platform.
- Managed: we run it, you sign in.
The default. Sign in with an email address or with Google and work in Lindi’s cloud. To connect a folder on your machine, use the Lindi Desktop app, which runs the Lindi bridge for you.
- Whitelabel: your platform.
Lindi under your brand and domain, with your design system built in — set up and run with you, the way Shamel runs for government teams in Saudi Arabia.
Lindi is new.
Here is what that means.
We have not completed a SOC 2 or ISO audit, and we are not going to put a badge on this page before we have. Everything above describes how the product is built — the roles, the sharing, the tokens, the switch. That is not the same as an audited guarantee about how it is run, and the two deserve different words.
If you are evaluating Lindi for a team where that distinction matters, ask us directly. You will get a straight answer about where we are, what is underway, and what we cannot do yet — and if the honest answer is “not yet, for you”, we would rather say so now than in month three.
Start coding while you design.
Lindi is free for its early users. Tell us about your team and we answer within 48 hours — yes or no.
- 1You connect your project
- 2We open your screens with you, on a call
- 3Your first push lands in your code