Jan

Install Jan on an air-gapped workstation.

Download a Llama-3-70B model.

Use the RAG extension to index 1,000+ case files.

Query the model for specific clauses and precedents offline.

Enable Jan's local API server.

Configure VS Code with the Continue.dev extension.

Set the API endpoint to http://localhost:1337.

Use local DeepSeek-Coder for autocomplete and debugging.

Setup Jan on a local server.

Write a script to pipe raw logs into Jan's local endpoint.

Use a prompt to identify and mask PII.

Output the sanitized data to a cloud storage bucket.

Load Jan onto a high-spec laptop.

Pre-download specialized scientific LLMs.

Perform data synthesis and thematic analysis of field notes on-site.

Maintain data integrity in remote areas.

Deploy Jan on a central internal Mac Studio.

Open the port to the internal network.

Direct employees to a simple web UI pointing to Jan's IP.

Scale AI access for $0 incremental cost.

Capture audio locally.

Transcribe using a local Whisper model.

Pipe the transcript into Jan using a local Llama model for clinical summarization.

Ensure no data leaves the clinic network.

Run Jan's Nitro engine across different GPU configurations.

Use the built-in performance logs to track tokens-per-second (TPS).

Compare GGUF vs TensorRT performance on the same hardware.

Export metrics for technical reports.