Explorar las categorías
Explora
Fiverr Pro
Español
$
USD
I will build a private rag
Ceonein Dedalus
Acerca de este Servicio
Your files stay on your machine. No subscription, no data sent anywhere.
I build private AI that reads your own documents PDFs, notes, files and answer question from them. Uses local open source models (Ollama), not cloud API. Nothing leave your computer.
Good for lawyer, clinic, consultant, any business that cannot legally send client data to OpenAI or similar. Also good if you just tired paying monthly AI bill forever.
How it work: your documents get split into chunks, turned into vectors, stored locally. When you ask something, system find exact relevant part of your file and give it to model as context. So answer come from your real document, not model just guessing from memory. This called RAG (Retrieval-Augmented Generation).
What you get: working local assistant that answer from your documents, tested before I deliver, full source code (nothing hidden), setup on your device or server you control.
What I don't do: fine-tuning or retraining model this is retrieval not weight training, I don't sell what I can't verify work.
- I'm new on Fiverr, message me before order if your case unusual, I tell you honest if it fit or not.
Conoce a Ceonein Dedalus
Ceonein Dedalus
Just doing it
- DeUzbekistán
- Miembro desdejun 2024
Idiomas
Inglés
I don't know why I'm doing this. They just told me I needed to do it and I agreed to it without thinking, but now I have no idea why I actually said yes to them.
Mi porfolio
FAQ
Is this the same as ChatGPT with my files uploaded?
No. ChatGPT sends your files to OpenAI's servers. This runs entirely on your own machine or a server you control — nothing leaves your device. That's the whole point.
Do I need a powerful computer for this?
A reasonably modern laptop works fine (8GB RAM minimum, 16GB is smoother). No GPU required, though it makes responses faster. I'll tell you honestly before we start if your hardware isn't enough.
Can it learn new information or be fine-tuned on my data?
Not in this gig — this is retrieval-based (RAG), meaning it searches your documents at question-time rather than retraining the model itself. That's actually better for most document use-cases: faster to set up, cheaper, and easy to update by just adding new files.
What file types can it read?
PDF, TXT, and Markdown by default. Other formats (DOCX, CSV) possible — message me your specific files before ordering so I can confirm.
What happens after delivery if something breaks?
Your package includes revisions (1/2/3 depending on tier) to fix issues with the agreed scope. I test before delivery, but if setup breaks on your specific machine, I'll help get it running.
