The AI Extension Builder lets you create a custom section in your GoodBarber app by describing it in natural language. The AI agent writes the code, plugs it into your app's APIs, shows the result in real time. No developer skills required.
A barrier comes down
You have a very specific section in mind for your app. A feature nobody has built before, because it speaks only to your community, your readers, or your subscribers.
Until now, two options. Either find an extension that comes close in our Extension Store, and live with the compromise. Or have the section coded by a developer, with the bill and the wait that come with it.
The AI Extension Builder adds a third path. You describe the section. The agent writes it, plugs it into your app, makes it immediately usable. Building custom sections leaves the developer's territory.
It's the most significant step we've taken on GoodBarber's AI roadmap this year.
What is the AI Extension Builder?
It's a new section type in your back-office, called Create with AI. You add it to your app like any other section. Inside, an AI assistant is waiting.
You describe what you want to see: "a weekly habit counter with a 7-day grid", "a tool that turns a list of ingredients into a shopping list", "a visual breathing coach with an animated bubble to guide inhale, hold, exhale". The agent generates the code, applies it, and shows you the result in real time.
You iterate by conversation. "Move the counter to the top right", "add a different color for successful days", "the validation button is too small". The section rewrites itself with every iteration. And if you can read a bit of code, you can open the generated file and edit it by hand — the agent has done the heavy lifting for you.
The case that struck me during the silent launch
We've made a habit, on key new product features, of rolling them out quietly to a subset of users before the public announcement. That's what we did with the AI Extension Builder over the past few weeks. Our idea was to observe what people build when they've received no instructions, no "suggested use case". To see what comes out, naturally.
One section, among those that were created, struck me in particular: a daily mood tracker.
The original prompt was three lines long. Something like: "A page where I can pick my mood for the day with an emoji, write a quick note, and see a chart of my last 7 days. If I'm a premium subscriber, I also want to see the last 30 days."
The agent delivered it in a single generation: a grid of clickable emojis (joyful, calm, irritable, sad, anxious), a short text field for the day's note, a validation button, a line chart for the current week. For premium subscribers, the chart automatically extended to 30 days.
Why did this section strike me? Because it illustrates exactly what I was hoping to see. A simple idea, immediately useful for the audience of a lifestyle or wellness app. And behind the apparent simplicity, four platform capabilities mobilized at the same time:
- User access — each member sees their own history, not someone else's.
- Persistent local storage — the recorded moods stay available offline, are saved on the device, survive an app restart.
- The memberships system — the length of history hinges on the premium subscription, on a single instruction from the creator.
- The app's global style — typography, colors, spacing, contrasts: everything matches the rest, automatically.
This would have meant several days of work for a competent developer. It's a few minutes here. And the section integrates like any other section of the app.
What the agent knows about your app
The mood tracker works because the agent doesn't code in a vacuum. It calls the platform's public APIs, the same ones our developers have used for years to build their custom sections and integrate them seamlessly with the rest of the app.
Its range of action goes well beyond the four building blocks mobilized in my example. The agent can query the phone's geolocation, open Apple Maps or Google Maps in a single call, trigger a system alert or native share, store data locally per user, read the subscription status, detect whether there's a network connection to adapt behavior. Anything that makes an advanced section work is within reach.
That's what separates a usable AI generation from a demo that looks great on social but can't go to production.
A few more examples
The mood tracker is one case among dozens. Many users have already built very interesting sections during the silent launch. We decided to draw on them to create a prompt library accessible directly in your back-office. Among others, you'll find:
- A Pomodoro timer that paces your work sessions
- A habit tracker to follow your weekly goals
- A parking memo that records your car's location with one tap and launches Apple/Google Maps to bring you back
- A recipe book that adjusts portions to the number of guests
Every creator brings their own ideas for their audience. With the AI Extension Builder, the path from idea to a section in the app has never been shorter.
AI at GoodBarber, on two axes
The AI Extension Builder fits into an AI trajectory we're building around two axes.
On one side, AI that helps you, the creator. In the back-office, Genius Palette (harmonious palettes in one click), the AI Assistant (writing, SEO optimization, multilingual translation), and now the AI Extension Builder. Outside the back-office, the MCP server and the 30 inspiration skills, published as open source, make your app drivable from any AI client compatible with the Model Context Protocol (Claude, Cursor, ChatGPT).
On the other, AI that serves your audience, inside the published app. The ChatGPT extension for a configurable generic assistant, and the RAG Chatbot for relevant answers grounded in your content (articles, events, map points).
Our ambition is to deploy AI everywhere it delivers a real service. To you, and to your audience.
Try the "Create with AI" section
The AI Extension Builder is available today on every account, under the Beta label. To try it, add a section to your app and pick the Create with AI type. Describe what you want. Five minutes is enough to see whether the experience holds up on your use case.
The prompt library will keep growing. The set of APIs accessible to the agent will keep expanding too, as we document new building blocks. If you build something out of the ordinary, write to us. Just like with the mood tracker, it's by watching what you do that we move the feature in the right direction.