Everything Ripul does, in your own app.

Calendars, bookings, payments, files, settings. Your app can do all of it, through buttons and screens your users have to learn. The SDK lets them ask instead.

“Schedule a stand-up every weekday at nine.” The agent understands, calls the functions your app already has, and reports back. You do not design a new screen for each new thing it can do.

Plain words as the interface

People say what they want. A job that took three screens becomes one sentence.

Nothing new to design

The agent is the interface. You describe a function once and the conversation is taken care of.

It combines things for you

Give it your calendar and your contacts and it can book lunch with Sarah, though you never wrote that.

Your code stays as it is

You add small wrappers around functions you already have. Your models and your logic do not change.

For each thing your app can do, you write a short description: its name, what it is for and what it needs. That is the new code. The SDK does the rest: showing the agent, carrying the conversation, calling your function and showing the result.

It scales. An app can offer a hundred functions without the agent losing its way, because you group them and the agent is shown the right group at the right moment.

People can talk to the agent in your app and have it talk back, with nothing extra for you to build.

The agent wears your app's colours and type, and you can change that look from a server, so a new theme reaches people without a new release.

This is the part that changes how you work. With the SDK in your app, an agent on your Mac can see your app's screen on a real phone, press its buttons, type into it and read its logs. It makes a change, installs it, tries it and tells you what happened.

It is how Ripul itself is built: the app is developed inside the app, by agents that check their own work on the device.

Sees the screen

The agent reads what is on screen as your app's own controls, not as a picture to guess at.

Presses and types

It taps the button a person would tap, and types where a person would type.

Reads the logs

When something breaks, it reads the error itself. Nobody pastes one into a chat.

Watch it live

See your app on the phone from your Mac or another phone while the agent works.

Ripul hosts your test builds. Publish one and your app shows its own list of builds, with history and release notes, and installs the one you pick. No storage or install page of your own to set up.

A customer taps Get Help in your app and reads you a six-digit code. You see their screen and can point at things on it. They sign in to nothing and install nothing.