Thoughts on this idea: A tool that turns a CSV + an idea into a working ML model (for non-coders)

Hey everyone, I've been kicking around this idea for a bit and wanted to get your thoughts. I'm building a tool called Da Vinci (haven't thought up a creative name yet).

The gist is, it's a super simple way for companies to build and use machine learning models. They'd just upload a CSV, tell it in plain English what they want to do, like "rank my leads" or "predict customer churn", and Da Vinci would train and spit out a tested, deployable model.

The output wouldn't be some complicated file they have to figure out what to do with. It would be a simple button in their Google Sheets or Airtable, or a small REST endpoint they can easily integrate into their existing workflows.
My whole angle is that a lot of businesses don't actually need these huge, expensive large language models for simple tasks like classifying support tickets or forecasting inventory. A smaller, cheaper, and more stable model that's wired directly into the tools they already use would be way more effective.

So under the hood, Da Vinci would be building something like a compact XGBoost or LightGBM model for tabular data, or a lightweight text model. It would also handle all the auto-validation and performance metrics.

I've been thinking about a go-to-market strategy focused on users of no-code/low-code platforms like Bubble, Airtable, and Zapier. I could offer them a "free demo" where I take their CSV and in return for their feedback, I'd give them the working model and a short loom video showing the results.

The thing is, this is all just an idea right now. I haven't had any customers or even really tested the waters to see if anyone would actually want this. It's essentially a "data-to-ML-model" pipeline run by AI agents, and I'm not totally sure if that's a clear value proposition for people.

What do you guys think? Is this a crazy idea? Is there a real need for something like this, or am I just building something cool for the sake of it? I'm really open to any and all feedback here.