Run it locally first. See how it behaves on real folders before you ask anyone for budget.
Free local AI for private documents
Open a folder. Ask a question. Check the passages behind the answer.
LocalRAG was built for normal working folders: OCR'd PDFs, scans, mixed formats, and documents that should stay on your machine.
The useful part is not the tone. It is the ability to open the passage behind the answer.
Private folders stay on your machine instead of starting their life in somebody else's cloud.
- Your files stay on your machine
- Answers come with source passages
- Built for messy working folders
- Models, embeddings, and roles stay visible
Made for the folders people usually hide in demos.
The product lives here. The code and releases live on GitHub.
Why LocalRAG
Not another shiny 'chat with files' demo.
Starts from your folder
Open a local folder and start working. No forced upload step and no default cloud hop.
Answers you can check
If you cannot open the passage behind the answer, the answer is asking for faith.
Handles real document mess
OCR, uneven filenames, mixed languages, scans, stray notes, and half-clean PDFs.
Controls stay visible
Models, embeddings, and response roles are part of the interface instead of being buried in side config.
What stays visible
The controls that matter are not buried in side menus.
Folder path, answer language, retrieval settings, roles, models, and retrieved passages stay visible before and after every question.
Local document folder
Answers with source passages
Ollama model selection
Embedding selection
Retrieval settings in the UI
Response roles by task
Screens from the running app
Screens from the running app, not a marketing mockup.
Everything below was captured from the current LocalRAG build.
Where it helps
Best for folders that are private, mixed, or easy to misread.
Private project archive
Specs, notes, meeting exports, and internal material that should stay with you.
Mixed-language research folder
One directory, many formats, uneven filenames, and still a need for a readable answer with evidence.
Source-sensitive work
Sometimes the exact fragment matters as much as the summary when you review a clause, note, or claim.
Corporate / Integrations
When LocalRAG has to work for a team, not just one person.
The corporate layer keeps the same core idea, but connects it to internal sources, existing workflows, and a controlled private environment.
Sources
File shares
Object storage
Knowledge bases
SQL content
LocalRAG corporate layer
Grounded answers, access control, connectors, evaluation
Same core workflow, plus the operating layer a real internal rollout needs.
Outputs
Internal APIs
Webhooks
Queues and events
Observability and evals
What changes in a team setup
You keep the same product logic, then add connectors, automation, access rules, observability, and the operating layer a real team needs.
Internal knowledge sources
File shares, object storage, knowledge bases, SQL-backed content, and other places where corporate information already lives.
Automation and integration
APIs, webhooks, and adapters so LocalRAG can sit inside existing workflows instead of becoming one more isolated tool.
Private deployment and operations
Private deployment, access control, observability, evaluation, and queue or event integrations for production environments.
Pricing
Start with the free build. Add the corporate layer when you need more control.
Free is enough for personal work, pilots, and evaluation. Corporate is for teams that need connectors, governance, and a proper operating layer around the same core product.
Free / Community
Run it locally, test it on real folders, and keep full control.
A good fit if you want to work with private documents on your own machine and see how the product behaves before adding more infrastructure.
- Local document workspace on your own machine
- Answer language configured inside the product
- Configurable models, embeddings, and response roles
- Free for personal use, evaluation, and pilots
Corporate
For teams that need integrations, private rollout, and operational control.
The same core product, expanded with connectors, governance, and the pieces needed for internal platforms.
- Connectors for file shares, knowledge bases, object storage, and internal content sources
- Extensible API, webhooks, and adapters for integration with corporate systems
- Database, queue, and event-pipeline integrations, including SQL platforms and Kafka flows
- Private deployment, access control, observability, and evaluation workflows for production teams
Next step
See the product here. See the code and release history in the repo.
This site shows the product. GitHub shows how it is built and how it evolves.