Vibe Coding with App Catalyst to Turn Prompts into Production-Ready Data Apps

If you’ve ever used an AI tool in your personal life and thought, “Why can’t work feel like this: magical, smooth, easy, and even fun?”—you’re not alone. Somehow, we’ve made building with data far harder and slower than it should be, especially for the people who deeply understand the business problems they’re trying to solve.  

This topic was at the heart of our Good Vibes x Domo event in January 2026. The goal wasn’t to show off a shiny demo. It was to explore a more practical idea: what does it actually take to go from a rough idea to a real, usable application when your data and your credibility are on the line.

To show how it’s done, we handed the mic to Florencia Silveira, a Domo data scientist who builds production solutions for real customers. What followed was a clear look at how she uses App Catalyst to turn “vibe coding”—describing your vision in natural language to AI tools, rather than writing code manually—into something you can actually use at work.

The problem isn’t talent. It’s smart people, stuck building the same things

Flo started with a frustration that goes way beyond data science. If you’ve ever rebuilt the same dashboard, app, or report, just for a different team, client, region, or quarter, you’ve felt it too and recognize the pattern.

Marketers see it when campaign performance apps all look the same, just pointed at different data. Finance teams feel it with forecast reviews and variance analysis. Data analysts know it when the structure stays constant, but the context keeps changing.

For Flo, this repetitive cycle showed up in model performance apps. As a data scientist supporting Domo customers, she often found herself building the same types of applications across different environments: forecast evaluations, regression metrics, and performance tracking.

The details change. The structure doesn’t. And yet, even when the data follows the same logic and with the same intent, everything still has to be rebuilt from scratch.

App Catalyst handles multi-DataSet apps without simplifying how the world really works

Real analytics work rarely runs on one neat table. Flo demonstrated this through a common example from her world: multivariate forecasting.

To evaluate model performance, you might need:

• One DataSet for model coefficients
• Another for predictions
• Separate tables for each model tested
• Additional DataSets to monitor live performance once the model is in production

It adds up fast. In practice, it can mean 10 or 12 Domo DataSets powering a single app.

Most tools force you to flatten reality to make it work. But Flo didn’t have to. With App Catalyst, she simply described the visuals she wanted, what each one was for, and which DataSet it should use. As long as the instructions are clear, App Catalyst can handle that kind of complexity instead of asking you to work around it.  

From prompt to production-ready app with App Catalyst

Flo began with a vision for her app, complete with titles, descriptions, purposes, and data sources. She gave enough context for the app she wanted.  

From there, App Catalyst built a real, multi-page app:

• A page for model selection metrics
• Another for live model performance

This wasn’t a mockup. App Catalyst turned Flo’s input into a functional application that was fully editable and extensible—something you could put in front of users.  

It’s an important point because a lot of AI tools are great at producing things that look impressive, but never get used. App Catalyst is designed differently: What you create shouldn’t be disposable. It should be ready to evolve.

The value of “vibe coding” really clicks in during editing. Once the app existed, Flo didn’t switch tools or rewrite anything. She just kept refining it through conversation.

Didn’t like the color scheme? No problem. She told App Catalyst the exact HEX codes she wanted. Prefer squares instead of dots for your data points? Done. That’s the beauty of vibe coding: you’re refining, not rebuilding. It lets you stay in your flow.

Building without a blueprint: How App Catalyst helps you start

Not everyone comes in with a perfectly detailed vision of the app they want to build, and that’s okay. Most work begins with a hunch, a half-baked idea, or a problem you feel more than you can describe.

To address this, Flo talked about elicitation and how good AI asks clarifying questions to gather context. App Catalyst goes a step further by turning your early, incomplete ideas into a proposed game plan you can actually use.

You can review how your app will be structured, adjust the approach, and iterate on the plan before anything is built. Once it feels right, App Catalyst executes. And even then, you can keep making changes afterward.

For business users especially, this removes a huge barrier. You don’t need to have the whole thing mapped out. To get started, you just need Domo and a vision.  

Why App Catalyst isn’t just another AI demo

Throughout the session, one thing kept coming up: There’s a big difference between experimenting with data and working with it responsibly. While most AI tools are built for the experimentation side of that divide, App Catalyst is deliberately designed for the other side.

Behind the scenes, this shows up in important ways that may not be obvious. App Catalyst works with the governed DataSets your teams already trust. It respects who should see what, so people don’t accidentally get access they shouldn’t. And it fits into existing data environments instead of bypassing them.  

For data teams, it means they maintain control even when someone starts creating. For business users, it means the freedom to explore and build without risk.  

Shifting from AI experimentation to scalable impact

The biggest takeaway from the session wasn’t a feature. It was a shift in how to think about building.

For a long time, teams have been told they have to choose: move fast or stay in control; be creative or be credible; explore new ideas or deliver finished products. But that distinction is breaking down.

We propose that “vibe coding” isn’t a novel fad but a professional way of building where curiosity and production coexist and work together. With App Catalyst, experimentation becomes the first step toward something real, not a detour to be cleaned up later.

If this sparked ideas for how you might build with your own data, the next step is simple.

Take a closer look at App Catalyst and see what vibe coding with your data could look like in your world.

Frequently asked questions