I’ve been quiet for a while. Those who’ve been following my writing know I usually can’t shut up about what’s happening in AI — the models, the tooling, the implications. But sometimes you go quiet because you’re building.
Today, I’m breaking that silence.
From Proto to Product
Back in June 2023, I offered free access to what was essentially a prototype — a rough-around-the-edges RAG system that let people connect their own data to large language models. It worked. People liked it. But it was held together with duct tape and optimism.
That prototype taught us a LOT. We learned what people actually wanted (simplicity), what they didn’t care about (flashy features), and where the real friction lived (getting data in, getting answers out). So we took those lessons and went heads-down for months, refining, rebuilding, and stress-testing with a handful of early clients who were willing to put up with our constant iteration.
Today, we’re officially launching Ragu.
What RAG Actually Is (And Why It Matters)
If you’ve been following the AI space, you’ve heard “RAG” thrown around constantly — Retrieval-Augmented Generation. The concept is straightforward: instead of relying solely on what a language model was trained on, you give it access to YOUR data at query time. It retrieves relevant chunks, then generates answers grounded in that context.
The problem isn’t the concept. The concept is pretty powerful. The problem is that most RAG implementations are either:
- Enterprise nightmares — six-month deployments, consultant armies, infrastructure that costs more than the insights it generates
- Developer toys — open-source frameworks that require you to understand vector databases, embedding models, chunking strategies, and a dozen other things before you can ask your first question
There’s a massive gap in the middle. People who have data, who have questions, and who don’t want to become infrastructure engineers to connect the two.
That’s what we built Ragu to solve.
Simple Is Hard
I’ve said this before about AI products generally — the competitive advantage isn’t the model. Everyone has access to the same foundation models now. The advantage is in the wrapper, the UX, the decisions you make so the user doesn’t have to.
With Ragu, the goal was radical simplicity. Upload your documents. Ask questions. Get answers grounded in YOUR data. No vector database configuration. No chunking parameter tuning. No prompt engineering required.
Behind the scenes, we’re making hundreds of decisions — how to chunk, how to embed, how to retrieve, how to construct the prompt, how to handle edge cases where the data doesn’t contain a clear answer. But the user shouldn’t have to think about any of that. They should just get good answers.
We’ve been testing this with about half a dozen early clients across different industries, and the feedback has been consistent: it just works. That’s the best compliment a platform can get.
The Nested Prompt Trick
One thing I’ve been playing with lately that’s adjacent to this — nested pre-drafted prompts. Basically, prompt templates that chain together, where each one feeds into the next. It’s a power-user shortcut that’s surprisingly effective for building repeatable workflows on top of RAG systems.
If you’ve ever found yourself typing the same complex prompt over and over with slight variations, this approach saves real time. I’ve used it a couple of times now and it’s clever — the kind of thing that seems obvious in retrospect but takes a while to click.
Free Access (Seriously)
Here’s the thing — I did this last June with the prototype, and I’m doing it again now. If you want to try Ragu, just fill out the form at ragu.ai/contact and reach out to me directly. We’ll give you free access.
The caveat: this holds up until your compute charges start to sting on our end. We’re not a charity, but we ARE confident enough in what we’ve built that we want people using it. The best way to improve a product is to get it in front of real users with real data and real questions.
What’s Next
The RAG space is moving fast. Every week there’s a new framework, a new approach to retrieval, a new embedding model that claims state-of-the-art. That’s exciting, but it also means most people can’t keep up — and they shouldn’t have to.
Our job is to keep up so you don’t have to. We’ll keep integrating the best approaches, optimizing the pipeline, and keeping the interface dead simple.
I’ve always believed the real AI revolution isn’t about the models themselves — it’s about making those models useful for people who have actual work to do. That’s what Ragu is. No PhD required. Just your data and your questions.
Give it a shot. I think you’ll be impressed.
