AI or BI: what do you prefer?

Quick survey of the community!

We’ve been talking a lot about the limitations of BI tools vs AI code generators & AI platforms for building apps.

I wanted to hear your thoughts: what’s your preferred workflow for building and sharing data insights?

Vote below and feel free to share your thoughts in the comments.

  • AI-powered coding (like Cursor, Windsurf, Copilot)
  • Data app platform with AI (like Dash)
  • Full-stack app development (coding manually)
  • Spreadsheets and BI reporting (Power BI, Tableau)
0 voters

your platform delivers more flexibility compared to the other solutions;
it is stable, easy to use and a good component for agile project planing compared to coding manually

4 Likes

The amount of community members that also use AI-powered coding tools (like Cursor or Windsurf) is much higher than I thought it was going to be.

When building Dash apps, would you say you use AI to solve bugs along the way or do you use AI to write the whole app for you?

I’m a beginner programmer. I present my reports using Dash by writing the code myself, and I rely a lot on ChatGPT to do it.

One challenge I often face is that many companies are already used to working with Power BI due to the Microsoft environment, and they prefer not to move away from it — which is understandable.

I use ChatGPT to guide me through how to do certain things (since I’m originally an economist, not a programmer). I also use it for repetitive tasks and to check for errors in my code.
In my experience, AI is a great help, but you can’t rely on it completely.

2 Likes

I have had the same experience @Jose_Ayala_Cambra. AI models like ChatGPT and Claude are powerful but we can’t relay on the completely for a final Dash app.

Have you tried other platforms or just ChatGPT?

1 Like

Since I work in VS Code, I also use the GitHub Copilot extension, although to a lesser extent. Even though both tools are developed by OpenAI, I used to think that ChatGPT, in addition to helping with code, also offers a broader perspective on the problems one is trying to address. That said, I tested GitHub Copilot Chat before answering this, and I can see that it actually responds quite well. Having support that goes beyond just coding can be especially valuable for those of us who are just starting out.

In this context, it’s important to mention that Dash is often the final step in a process. The use of ChatGPT, on the other hand, tends to come from the very beginning. For example, if I’m presenting a descriptive analysis of customers that includes segmentation, there’s already prior work before using Dash, where one chooses the platform to rely on for presenting the results. By the time I use Dash for the final report, the supporting tools have already been chosen.

2 Likes

Thanks to everyone who weighed in on our recent poll about how you’re building and sharing data insights. Sharing some thoughts on the results:

The top response was full-stack app development, with 51% of respondents still coding manually. That tracks with what we’ve seen that experienced data scientists and developers still want full control when building advanced data apps.

But it looks like there’s a shift happening with 45% saying they’re using AI-powered coding tools such as Cursor or Copilot. And I think it’s mainly because AI is helping people generate boilerplate code.

Another 40% chose AI-enabled data app platforms like Dash. As my colleagues like to say, these sit in a sweet spot: you still write code, but you get help accelerating deployment, UI, and collaboration.

Spreadsheets and BI: only 22% said they prefer that route, which is probably because traditional BI tools weren’t designed for this kind of customization or AI integrations.

Thanks everyone for participating in this poll! Your votes made this a valuable piece of insight for our community. If you’re interested in this topic and want to read further, my colleagues have published a new paper on how AI is reshaping the analytics stack. Give it a read: A smarter, faster way to build what BI never could

3 Likes