Alpine mountain ridge at dusk

Manage the build, not the chat.

Bennch is the workspace for software built with AI. Your coding agent connects to it over MCP, builds from a clear plan of cards and reports back, so you can see what’s planned, what’s being built, what needs your decision and what’s actually been checked.

Your AI writes the code. Bennch shows what’s planned, what’s being built, what needs you and what’s been tested.

Live now with Claude Code. Codex and more coming soon.

Works with
Claude Code
Codex (soon)
GitHub
GitLab (soon)
Sentry (soon)
Bennch Overview for the example project Ledgerly: in Build with 19 of 38 cards done, two AI sessions working, what’s waiting on you and where to try the latest

Ledgerly, an example project in Bennch: the next step, what needs you and where the build stands.

AI writes the code. Bennch keeps the project on track.

Building fast is the easy part now. Knowing where the project stands is what gets hard.

01 — Lost context

The whole project, every session.

The plan, your decisions and design links live with the project, so a new session never starts from zero.

02 — Assumptions

Your AI builds what you approved.

Extra ideas come back as proposals. Choices that are yours come back as questions.

03 — “Done”

Done means you’ve checked it.

Not when the AI says so. Each card shows its commits and tests, and only you can accept it.

How it works

Approve the cards. Your AI builds them in order.

Every piece of work is a card. Your AI takes them in order, in the same Claude Code session, and hands each one back with proof.

Up next 4

Your AI works down the list

LD-24Frontend

Turn an accepted quote into an invoice

LD-25Frontend

Attach photos of the finished job

LD-31Frontend

Mark an invoice as paid in cash

LD-32UX/UI

Monthly income summary

Your AI is building 1

One at a time, in order

LD-23UX/UI

Today screen: jobs and unpaid invoices

iOSAndroid

2 commits so far

Tests passing

Next up: LD-24, in the same Claude Code session. Nothing for you to start.

Your check 2

Built. Only you can finish these

LD-22Security

Sign in with an email link

iOSAndroid

4 commits · tests passed

You
Mark done
LD-21Backend

Remind clients when an invoice is overdue

2 commits · tests passed

Swipe to see more →

1You approve the cards

What each is for and how you’ll know it works. Unapproved cards can’t be started.

2Your AI builds them in a row

One card, or as many as you like. Your AI finishes one, starts the next, and only stops when something needs you.

3You check the proof

Try it on your own devices. A failed step goes back to its card with your note.

The whole project

More than a task board.

The plan, the decisions, the code and the tests stay connected from the first card to launch.

A plan for every part

A plan that covers the whole product.

Take Ledgerly, an invoicing app for tradespeople. It has a phone app, a page where clients pay, and a backend. Each card belongs to one of those parts, sits on a board for its kind of work, such as design, backend or security, and says where it runs: iOS, Android or web.

See the whole product at once, or just the phone app’s next few cards, and have your AI work through them in order.

Every card belongs to a part of your product, sits on a board for its kind of work and says where it runs.

  • LD-24Frontend
    Turn an accepted quote into an invoiceLedgerly app · iOS, Android
  • LD-27Payments
    Refund a card paymentClient pay page · Web
  • LD-28Data
    Keep invoices for six yearsBackend
The Bennch board for Ledgerly: tabs for each kind of work, then cards in Up next, being built, Your check and Done, each showing its part of the product and where it runs
Decisions: open questions about the card payment provider, ways to pay and where PDFs are kept, each with its options and why it matters
Decisions

Decisions stay with the work.

When a choice is yours, such as how data is stored, payments or security, your AI asks with options and a suggestion instead of deciding. Answer in the chat or in Bennch.

Your answers stay on record for every later session, and cards link to where their designs live, like a Figma file.

When a choice is yours, your AI asks, with options and a suggestion. Your answer stays on record for every later session.

Progress across sessions

Pick up where you left off.

The Overview shows the next step and what’s waiting on you. If a session stops halfway, the next one picks that card up first. Run several AI sessions at once and each card shows which one is building it.

If a session stops halfway, the next one picks that card up first. Run several at once and see what each is doing.

The Overview: the stage tracker, a session that stopped partway with its commits saved and a Continue button, and two sessions working now, each on its own card
A card in Your check: 4 commits with their sizes, tests for iOS and shared logic, a note that Android has no passing run yet, and buttons to accept it or send it back
Tests and evidence

Evidence attached to what was built.

Each card shows its commits, the tests the AI reported, your CI result and which platforms were covered. Work that grows far beyond its card is flagged.

Each card shows its commits and tests, and what hasn’t been tested yet.

To do

The jobs only you can do.

Your AI can’t create your accounts, set up your server or add your API keys. When it needs one of those, it adds it to your To do list with short steps and shows which cards it’s holding up.

Once you’ve done it, your AI tests it where it can. Your keys stay out of Bennch.

Your AI lists what only you can do, like creating accounts or adding API keys, with short steps. Your keys stay out of Bennch.

A To do list: set up the email service, suggested by Claude; point your domain at the server, where Claude tested it and it still points to the old address; and create your Stripe account, which is holding up card LD-27
Testing

Test it yourself, one step at a time.

Each journey through your product gets a step-by-step guide, linked to the cards it covers. Follow it on your own computer or phone, and one run can accept several cards.

A step-by-step guide on your own computer and phone. One run can accept several cards.

A test step: step 3 of 4, close the app and open it again, you should see you’re still signed in, this checks card LD-22, and Is that what you see? with Yes and No buttons
Open the app: a QR code to scan that opens the TestFlight build on your iPhoneWhere to test: an Expo preview link for Android, a staging site for the web and a TestFlight link for iPhone
Swipe for the QR code and test links →
All tests: a full run and four journeys, with the result on web, iPhone and Android beside your AI’s own tests, including one failing

1Open it where it really runs

Staging, Expo or TestFlight. Scan the code and it opens on your phone.

2Follow the steps

Each step says what to do, what you should see and which card it checks. A problem goes back to that card with your note.

3See what’s been tested

Every journey on every device, beside the results from your AI’s own tests.

Start where you are

Start from an idea, or from code you already have.

Tell Bennch which it is. An idea gets planned with you; existing code gets mapped: what’s built, and what’s left.

A new idea

Plan it together

Your AI asks what you’re building, who it’s for and what the first version must do. It plays back a short brief, then proposes cards for you to approve.

Your AI: “What are we building?”

Existing code

Map what’s there

Your AI reads the code and proposes cards for what’s built and what’s left. Built work comes to you to check in one go; new cards are only for what’s missing.

Your AI: “This folder has code, so I’ll map what’s built and what’s left.”

Setup

Up and running in a few minutes.

01

Join the waitlist

We’ll email your invite. A short practice project, about 20 minutes, shows you how it all works.

We’ll email your invite, with a short practice project to learn on.

02

Add Bennch to your AI

Add it as a custom connector in Claude (Customize → Connectors) with app.bennch.co.uk/mcp. It connects over MCP and covers all your projects. No keys to copy.

Add it as a custom connector in Claude with app.bennch.co.uk/mcp. No keys to copy.

03

Ask your AI to start

Paste the start prompt from Bennch. Your AI plans or maps the project with you; approve the cards, press Build, and it gets to work.

Paste the start prompt. Your AI plans your idea or maps your code, then you approve the cards.

Security

What Bennch enforces.

  • Bennch never stores your code

    It keeps your plan and what your AI reports: commits, tests and results.

  • Only you approve and accept

    Your AI’s connection has no tool to approve cards or mark work done.

  • Secrets are refused

    Values that look like API keys or tokens are refused on cards, decisions and handovers.

  • Approved work only

    Bennch refuses to start a card you haven’t approved.

Bennch never stores your code

It keeps your plan and what your AI reports: commits, tests and results.

Only you approve and accept

Your AI’s connection has no tool to approve cards or mark work done.

Secrets are refused

Values that look like API keys or tokens are refused on cards, decisions and handovers.

Approved work only

Bennch refuses to start a card you haven’t approved.

Integrations

Works with your tools.

  • Claude CodeLive

    Builds from your approved cards over MCP. Bennch runs no AI of its own.

  • GitHubLive

    Commits that name a card appear on it, with your Actions result. Read-only.

  • CodexComing soon

    The same cards and checks for OpenAI’s Codex.

  • GitLab and SentryComing soon

    Pipelines on each card, and new errors on the card that shipped them.

Claude CodeLive

Builds from your approved cards over MCP. Bennch runs no AI of its own.

GitHubLive

Commits that name a card appear on it, with your Actions result. Read-only.

CodexComing soon

The same cards and checks for OpenAI’s Codex.

GitLab and SentryComing soon

Pipelines on each card, and new errors on the card that shipped them.

FAQ

Questions.

Which AI coding tools work with Bennch?

Claude Code works today, and Codex and other coding agents are coming. Your AI does the building; Bennch holds the plan, the decisions and the evidence.

Does Bennch run AI?

No. Everything the AI does happens in your own coding tool, under your own account.

Can Bennch stop the AI going off-plan?

Bennch blocks what goes through it: your AI can’t start an unapproved card, approve anything or mark work done. It can’t control what the AI writes in your files, so with GitHub connected, commits are linked to cards, unusually large changes are flagged, and nothing counts as done until it’s been checked.

Can my AI work through several cards?

Yes. Press Build, pick a board or part of the product, and choose one card or several in a row. You can also run several AI sessions at once.

Know where your build stands.

Early access is opening to founders, makers and innovators building with AI, starting with Claude Code.

Already invited? Sign in · Privacy