Private RAG search
Ask your documents as you would ask an expert colleague. The AI answers citing the document and page, and respects who can see what.
A private RAG is an AI search tool that answers from your internal documents and cites the source, without them leaving your server. For law firms, clinics and manufacturers with a lot of documentation. It is up and running in 3 to 5 weeks, with group permissions and quality testing before handover, in Spanish, Catalan and English.
Contracts, briefs, protocols, manuals, minutes, internal regulations. Any business with a few years of history builds up thousands of documents and, when someone needs an answer, they search by hand or ask the colleague who knows. The temptation is to upload everything to a cloud ChatGPT, but those are exactly the documents that shouldn’t leave: personal data protected by the GDPR, client information under professional secrecy, industrial know-how or case files. And if you then offer that assistant to your clients, the EU AI Act (Art. 50) also requires you to tell them they are talking to an AI.
A private RAG solves both problems: the AI answers from your documents, cites the source and everything lives on your server. In practice, it is a private company ChatGPT that knows your archive.
If you don’t have a server yet, we start with local AI for business and the RAG is built on top.
If you have a few dozen documents, a good folder structure and your system’s built-in search are enough. If your documents already live in a tool with a good search of its own, you may only need to connect it. If they are public (for example, legislation published in Spain’s Official State Gazette, the BOE), a cloud service can handle it with no risk to your data. And if the documentation is disorganised, out of date or contradictory, the AI will inherit that mess: sometimes the first step is to put things in order, and we’ll tell you so.
Our own agents, in our Barcelona lab, store and retrieve information with Qdrant, embeddings and a reranker running on our GPUs, without touching the cloud. The agent that manages a family estate in the Penedès also keeps its memory in Qdrant. It is the same component we install in your business, proven in daily use and not just in a demo.
An assistant that knows your case files and templates, runs inside the firm and never exposes a single piece of client data.
Less time typing and more time with the patient: transcription and summaries that stay inside the clinic.
Your technical know-how, searchable in seconds and without leaving the company network.
Tutors and assistants that teach with your method and your syllabus, without sending student data to the cloud.
AI in Catalan for administrative work, installed inside your security perimeter.
Between 3 and 5 weeks, depending on the volume of documents, the number of sources and whether group permissions need to be managed. If you don’t have local AI installed yet, we set up the server first and the RAG goes on top.
It is a technique that, before answering, searches your documents for the relevant passages and passes them to the model as context. That way the answer is based on your information rather than on what the model remembers, and you can check the source.
It can get things wrong, but far less often than a generic chatbot, because it answers from real passages and cites where they come from. We configure it to say it can’t find the answer when there is no supporting document, and we measure accuracy before handover.
Yes. We use multilingual embedding models that work well in Spanish, Catalan and English, and you can ask in one language about documents written in another, which is very common in Barcelona businesses.
A private RAG is an AI search tool that answers from your internal documents and cites the source, without them leaving your server. For law firms, clinics and manufacturers with a lot of documentation. It is up and running in 3 to 5 weeks, with group permissions and quality testing before handover, in Spanish, Catalan and English.