Privacy-first by default
The working point is a local documents folder on your own machine.
Local-first AI for private documents
Your documents stay local. The answer still needs to prove itself.
LocalRAG is built for the folders people actually keep: PDFs after OCR, mixed formats, multiple languages, and notes that do not belong in a public cloud.
- Local folders stay on your machine
- Answer and source context stay together
- Interface and answer language are separate
- Models, embeddings, and roles are adjustable
Made for mixed PDFs, OCR noise, and multilingual folders.
Public site here. Source and release context on GitHub.
Why LocalRAG
Built for messy real folders, not clean benchmark demos.
Grounded instead of smooth-talked
The answer is more useful when you can inspect the source context instead of trusting a polished paragraph.
Multilingual where it matters
UI language and answer language are separate controls. That matters when the folder itself is mixed-language.
Operator control stays visible
Models, embeddings, and response roles stay in the interface instead of being hidden in setup files.
What the UI already shows
You can see the operational parts, not just the chat box.
The current build makes folder path, answer language, retrieval behavior, roles, models, and retrieved context visible in the UI.
Local document folder
Source-grounded answers
Configurable Ollama models
Embeddings selection
Multilingual UI
Role-based responses
Interface languages
The same five interface languages as in LocalRAG.
The interface currently ships with English, Russian, Dutch, Chinese, and Hebrew. Inside the app, answer language is configured separately.
English
Russian
Dutch
Chinese
Hebrew
Screens from the running app
These are product screenshots, not designed mockups.
Everything in this gallery comes from the current LocalRAG build.
Where it helps
It becomes more useful in three very ordinary moments.
Private project archives
Notes, specs, meeting exports, and internal docs that should stay on your own machine.
Mixed-language folders
One folder, several languages, inconsistent filenames, and still a need for a readable answer with evidence.
Source-sensitive decisions
The supporting fragment matters as much as the answer when you are checking a clause, note, or claim.
Pricing
Start with the free build. Add corporate capabilities when the environment gets bigger.
LocalRAG is free for individual use, evaluation, and local-first workflows. Corporate work starts when the team needs connectors, enterprise integrations, and operational guarantees around a private deployment.
Free / Community
Use it locally, learn it deeply, keep it free.
The public version is designed for personal workflows, technical evaluation, and community-driven use around private folders.
- Local-first document workspace
- Multilingual UI with separate answer language
- Configurable models, embeddings, and response roles
- Free for personal use, evaluation, and community contribution
Corporate
For teams that need integrations, rollout control, and enterprise plumbing.
Corporate deployments can extend the same local-first core with enterprise connectors, API surfaces, and integration patterns for internal platforms.
- Connectors for knowledge bases, file shares, object storage, and internal content sources
- Extensible API, webhooks, and adapters for integration with corporate systems
- Database and event-pipeline integrations, including SQL platforms and Kafka-based flows
- Private deployment patterns, access control, observability, and evaluation workflows for production teams
Start here
Inspect the product here. Inspect the implementation in the repo.
The site is the product-facing layer. GitHub is where the code, release context, and technical history live.