A visual system, not a template
Tokens for colour, type and spacing are set once and used on every screen, so the product reads as one thing.
ML AI Code
Build anything. Infinitely.
Describe it once. A Frontend engineer sets the design, then ML AI Code writes, tests and ships the software around it. Every result is checked before it is called done.
Design, led by Frontend
Design comes first. Then the code.
In Auto, a Frontend engineer owns the look of everything with a screen. It writes a short design brief, sets one visual system, uses real copy and imagery, then checks the result in a browser at phone and desktop sizes. It is part of every task, at no extra cost.
Colour
One accent, calm neutrals, contrast checked
TypeAa
One display face, one text face, five sizes
Spacing
An 8 point rhythm, two radii, two shadows
Decks you edit in place
Sixteen by nine, premium layouts. Change words, images and order on the right, present full screen, export PDF or PowerPoint.
Images when they help
Hero and section images are made by the built-in image tool in one art direction. Icons, charts and patterns stay crisp in code.
BriefWho it is for, the job of each screen, the mood.
BuildEvery state: loading, empty, error, hover, focus.
CheckOpened in a browser at phone and desktop sizes.
Included with every task. No switch to turn on, no extra credits.
The problem
Coding agents are impressive for ten minutes.
Then you hand one a real task in a real repository, and the same six failures show up in tool after tool. They are rarely about how clever the model is. They are about everything around it.
- 01
It drifts from what you asked.
Asked to fix one bug, it rewrites three files nobody mentioned, and calls half a job finished.
- 02
It loses the thread.
Long tasks outgrow its memory. Your original request and your corrections are the first things to fall out.
- 03
It breaks what already worked.
One bad edit leaves a file that no longer parses, and there is no clean way back.
- 04
It says done when it is not.
“All tests pass”, when no tests ran, a failure was hidden in a summary, or the code changed after the check.
- 05
Only a prompt fences it in.
A line of instructions is all that stands between the agent, your keys, a force push and production.
- 06
It forgets, and it stops.
Close the laptop or lose the connection, and the work, and everything you decided, is gone.
The answer
Six failures. Six systems. Built into the runtime, not hoped for in a prompt.
Every one of those failures is an engineering problem, so we engineered the answer. Each system is ordinary, inspectable code with its own tests, sitting between your request and the result.
It drifts from what you asked.
01Task contract
Before the first edit, the request becomes a written contract: what to deliver, which files are in scope and which checks must pass. Progress is derived from that contract, and a task cannot finish while a deliverable is open.
It loses the thread.
02Context management
Long tasks outgrow any working memory. Older tool output is summarised at a fixed threshold while the original request, every correction and the most recent steps stay word for word. Nothing is deleted: the full history can be read back.
It breaks what already worked.
03Guarded edits
Every change is checked as it is written. An edit that breaks a file which parsed a moment ago is refused, a checkpoint is taken before each step, and parallel helpers work in their own copies so nothing collides.
It says done when it is not.
04Independent verification
"Done" is treated as a claim. The run records which commands ran, how many tests they covered and whether the code changed afterwards. Claims without matching evidence get one correction round, then are reported as unproven.
Only a prompt fences it in.
05Safety policy
Each command is classified before it runs. A short list is always refused, a repository file sets the rules for that project, and a team can only tighten those rules, never loosen them.
It forgets, and it stops.
06Continuity
Work survives interruptions. Actions with side effects are journaled before they start, decisions and open questions carry to the next run, and specialists can be resumed with what they already found.
Measured, not claimed
We test the whole system on real repositories.
Public tasks from widely used open-source projects, graded outside the agent on a clean copy. We publish first attempts and later repairs as separate numbers, and we say plainly what the evidence does not show.
19/30repository issues accepted on the first attempt
25/40hard multi-file tasks accepted on the first attempt
69/70tasks with an accepted patch across retained attempts
4ways a false “done” is refused
Development evaluations on selected public tasks, not an official score. Method and limits.
Vision
Anyone who can describe software should be able to ship it, and trust it.
Mission
Build the engineering that makes coding agents dependable: they stay in scope, check their own work, and tell you plainly what they could not prove.
- Proof over promisesEvery “done” comes with the commands, tests and diff behind it.
- People decide, agents doYou approve the plan and anything risky. It does the rest.
- Honest about limitsWhat could not be verified is reported as unproven, never hidden.
How it works
Hand it an issue. Get back working code.
It reproduces the problem, edits the code, runs your tests and starts a live preview, while you focus on the decisions.
01
Plan
Reads the repository and writes a task list you can follow.
02
Edit
Changes only what the plan names, with a checkpoint first.
03
Run
Executes commands and tests on its own machine.
04
Show
Opens a live preview and a diff you can review or turn into a PR.
Asks before it edits
You pick the tasks. It keeps to them.
In Guided mode the agent shows its plan first. Untick anything you don’t want; the rest stay visible as cancelled, and every command waits for your go-ahead.
Review the planPick what to do
✓Guard the Pay button against a second submit
✓Send an idempotency key with each charge
Rewrite the checkout page styles
DismissDo 2 selected tasks
Works where you are
Start on your phone. Finish on your laptop.
Each project gets its own isolated cloud computer, or keep everything on your own machine. One account across devices.
Phone
Full workspace in the mobile browser and the iPhone app, built for one hand.
Browser
Projects, previews, diffs and terminal side by side.
Laptop
Run the agent locally from the command line. Files never leave your computer.
Pricing
Start free. Upgrade when it’s doing real work.
Every plan has the full agent. Plans differ by monthly credits, active tasks and team features.
Free
$0forever
Trying the product
- 10 credits / month
- 25 welcome credits, once
- 1 builder
- 1 active cloud task
- Help center
Starter
$9/ month
Occasional coding
- 150 credits / month
- 1 builder
- 1 active cloud task
- Email support
Most popular
Pro
$25/ month
Regular builders
- 600 credits / month
- 1 builder
- 2 active cloud tasks
- Priority email
Team
$99/ month
Teams building together
- 2,400 shared credits / month
- 3 builders, +$25 per extra builder (adds 600 credits)
- 3 active cloud tasks, +1 per extra builder (max 6)
- Assign, claim and hand off tasks
- Shared projects, roles and free viewers
- Shared credit pool and member limits
- Priority admin support
Top-ups: $10 for 250, $25 for 625, $50 for 1,250, $100 for 2,500 credits. Purchased credits never expire. Tasks stop at zero; auto-recharge is off unless you turn it on. App hosting is billed through your own host account. Payments by Stripe.
Questions
Frequently asked
What is ML AI Code?
A coding agent in your browser. It takes a task, writes a plan, edits code on its own cloud machine, runs your tests and previews the app, then reports exactly what changed.
Does it change things without asking?
Not in Guided mode. It shows the task list first and you pick what to do. Commands wait for your approval, or you can approve the rest of a task in one tap. Auto mode is there when you want it.
What are credits?
Credits cover AI work, cloud computer time and paid tools such as voice transcription. A small fix uses a few credits; a large feature with tests can use 30 or more. The default Pixis model uses the fewest; other models use more, and the model menu shows how many times more. Every task shows an estimate first.
What happens when I run out?
Tasks stop at zero, so you never get a surprise bill. Buy a top-up at $0.04 per credit (purchased credits never expire), upgrade, or turn on auto-recharge with a monthly cap you set. Monthly plan credits reset each cycle.
What does Team add?
Shared projects, assigning and claiming tasks, handoffs, continuing a teammate’s task, a task board, roles with review and release controls, free viewers, one bill with a shared credit pool and per-member limits.
Where does my code run?
Each project gets its own isolated cloud computer with the common languages, databases, Docker and cloud CLIs preinstalled. You can also run the agent on your own laptop, where files never leave it.
Can I cancel anytime?
Yes. Manage or cancel in the Stripe billing portal from Settings. Your plan stays active until the end of the period.
The future of software starts with one sentence.
Write yours. Your first machine is ready in seconds.