Cartesiano.ai

Just launched on Producthunt! https://www.producthunt.com/products/cartesiano-ai?launch=cartesiano-ai

A few years ago, I remember getting a LinkedIn message from Tally's founder Marie Martens. The reason I remember (and the reason she messaged me in the first place!) is because on that day, Tally was launched, and I had also just launched a small side-project at the time, so she was reaching out to gather feedback.

I remember thinking "Why would anyone build a form builder?" It felt like there were already hundreds of options online. It was a question I’d asked myself about a lot of product ideas, and this felt like the most obvious example. So, honestly, I figured Tally would be another blip on the radar. I couldn't have been more wrong of course!

Watching Tally's journey has been incredibly inspiring. They took a familiar concept (form building), and executed it beautifully. They prioritized a delightful user experience, focused on ease of use, and gradually built a passionate community. They didn't chase unicorns, they just built a really good product and let it grow organically.

I’ve been following them ever since, and it has caused me to shift how I approach side-projects. I used to get paralyzed by the pursuit of completely new ideas. If I couldn't come up with something unique, I'd dismiss the idea as “not worth it.” That constant search for originality led to a lot of discarded projects and self-doubt.

Now (thanks to Tally), I view things through a different lens. It’s not about finding a totally new market. It’s about recognizing when a market exists, understanding why people are underserved, and asking myself, “What can I do differently? How can I offer something better or cater to a specific niche?”

That’s precisely what drove me to build Cartesiano.ai, a service to track your brand's presence across LLMs. The problem is, existing solutions are often too expensive, or targetting corporate clients.

I originally thought the idea behind Cartesiano.ai was relatively novel. But, a quick Google search proved me wrong! It was a moment of "Oh, okay, that's why there's a market for this." It wasn’t a discouraging moment, though. Instead, it fueled my determination and drive to get a first MVP version out as quickly as possible. It confirmed that people were actively searching for a solution, and I felt even more motivated to create something that could truly resonate with them.

It’s a humbling experience, building in public, especially when you realize you're not pioneering a completely new frontier. But Tally taught me that execution, community, and relentless focus on user value can be just as powerful than pure originality. The journey is just beginning for Cartesiano.ai, and I'm excited to share it with you all.

I'm very happy to launch Cartesiano.ai, a service which allows companies to track their presence across different LLMs.

There's been a few products popping up in this space the past 12 months, so the idea itself is not new, but there isn't a clear market winner yet, and most have either been backed heavily by VCs and/or are meant for more corporate users, so I'm betting on there being space for more competitors!

The Problem

Say for example you are Nike and notice that some of your inbound traffic is no longer coming from just Google, but ChatGPT. However, ChatGPT and other LLMs don't currently offer any form of analytics into what users are asking when it recommends your company or products.

To solve this, we try to think of what our target users might be asking (their prompts). So an LLM recommending Nike for example, might be the result of a user asking "What are the best marathon". With Cartesiano.ai, you can track these prompts across different LLMs, and track, over time, how often your brand/product is being mentioned, its ranking compared to other competitors, and its sentiment. It can also automatically detect new competitors you may not be aware of.

Why did I build it?

I built it because I was both looking for a side-project idea where I would have to work directly with LLMs, and one where there was already proof that a market need existed for it.

Tech-stack

From the tech-side, I've built this using:

• Ruby on Rails 8 for the main user-facing web app.

• Python + FastAPI + LiteLLM for the LLM processor application, which basically handles running the prompts through the different LLM providers (ChatGPT, Gemini, etc.), parsing the responses, normalizing the data, detecting sources, competitors, etc.

• PostgreSQL as the main database

• Hetzner for hosting (2 servers, one for the Rails and Python apps, another dedicated one for the self-hosted PostgreSQL database)

If anyone is reading this and interested in more of the tech-side, let me know and I can write a more in-depth post on it!