About LearnByAi
AI document analysis built for privacy
LearnByAi is a self-hosted Retrieval-Augmented Generation platform. Upload any document and ask questions about it in plain English. No subscriptions and no forced sign-up.
- No account required
- Pay-as-you-go
- Delete anytime
- 8 document types
Six principles behind every product decision
These are not aspirations. They are the concrete commitments that shaped how LearnByAi was built.
Anonymous-first access
No sign-up wall. A unique browser token creates an isolated workspace the moment you land. Your documents and chat history stay private to your device until you choose to create an account.
Pay only for what you use
No subscriptions, no seat licences. Top up your wallet and spend only when you trigger a metered action. Funds never expire, and the free tier gives you a meaningful allowance before you ever need to pay.
Document isolation by design
Every document is tied to its owner at the vector layer, and every query is filtered by that owner, so your content is not returned to other users.
One interface, every document type
Legal contracts, medical reports, financial statements, HR files, construction scopes, academic papers. The same clean chat interface handles them all through a unified RAG pipeline with specialized prompts per document type.
Your data, your control
Documents and conversations are sent to OpenAI only to answer your questions, through its commercial API. OpenAI's API data-handling terms apply to that data.
Deletion on request
Delete a document and it is removed from file storage, the vector index, and the database. Where backup copies exist, they are cycled out on their normal schedule. Request account deletion and your data is removed with it.
One interface for every document type
LearnByAi routes each upload through a specialized AI pipeline with tailored prompts, retrieval parameters, and quick-action sets matched to the document type.
- General document analysis: Any PDF, DOCX, or text file
- Legal contract review: Contracts, agreements, legal opinions
- Medical report analysis: Health records, discharge summaries, lab results
- Academic study assistant: Research papers, lecture notes, coursework
- Financial document analysis: Statements, audits, investor documents
- HR and employment documents: Contracts, policies, job descriptions
- Construction and project documents: Scopes of work, blueprints, contracts
From upload to answer: how it works
LearnByAi uses Retrieval-Augmented Generation (RAG) to answer questions about your documents. Here is what happens behind the scenes.
- You upload a document. Your file is transferred over an encrypted connection to isolated cloud storage. Text is extracted from the document, including OCR if needed for scanned pages.
- The text is embedded and indexed. The extracted text is split into chunks, converted into vector embeddings using OpenAI's embedding model, and stored in your document type's collection in Qdrant.
- You ask a question. Your question is also converted into a vector embedding. LearnByAi retrieves only the chunks from your document that are most relevant to your question.
- The AI writes an answer from those passages. The retrieved passages and your question are sent to OpenAI's chat API. The response is based on the passages retrieved from your document, so you can check it against the original.
What powers LearnByAi
Every tool in this stack was chosen for isolation, reliability, and data sovereignty. The core components are self-hosted, giving us direct control over where your data lives.
AI and Language: OpenAI
Embedding generation and contextual chat responses through commercial API routes. OpenAI's API data-handling terms apply to data submitted through the API.
Vector Storage: Qdrant (self-hosted)
Vector database running on our own infrastructure. Document embeddings are stored in separate collections per document type, filtered by owner at query time.
Database and Auth: Supabase (self-hosted)
Relational records, session management, and user authentication with hashed passwords and secure session tokens.
File Storage: Cloudinary
Asset storage. Files are accessed only via signed, expiring URLs generated by our API.
Infrastructure: Oracle Cloud
Application servers and API layer running on Oracle Cloud Infrastructure, managed directly by the LearnByAi team.
WhatsApp Interface: WhatsApp Business API
Document upload and AI chat accessible directly from WhatsApp, built on Meta's WhatsApp Business Platform.
Payments: Polar
Card payments for wallet top-ups are processed on Polar's hosted checkout page. LearnByAi does not receive or store your full card number.
What LearnByAi is and isn't
What it is
- A tool for understanding document content faster
- A private, session-isolated workspace for your files
- A starting point for research, review, and analysis
- Honest about what the AI can and cannot do
What it isn't
- A replacement for a lawyer, doctor, or financial advisor
- A guaranteed source of factually accurate outputs
- SOC 2 or ISO 27001 certified (not yet)
Upload a document and see it work
No account required. Your document stays private to your session, and the first questions are free.