Privacy Mode keeps Agent work on your machine — local models + Local Semantic Index.
What it is
Privacy Mode is a first-class switch in Elido Cortex. When it is on, Cortex Agent uses Elido Local Engine and on-machine models. Cloud providers are hidden from the model picker. There is no silent fallback to cloud GPU.
Local Semantic Index
Alongside Privacy Mode, Local Semantic Index chunks, embeds, and searches your codebase on-device. Embeddings stay on your machine so codebase_search does not upload your source to build the index.
How to turn it on
- Open Settings → Privacy (or the Privacy Mode control in Cortex Agent chat).
- Start Elido Local Engine and pick a local model that fits your GPU / machine.
- Use Cortex Agent as usual — prompts and code context stay with local inference while Privacy Mode is on.
What stays local
- Agent chat and completions via Local Engine
- Semantic index chunks and embeddings
- Your workspace source files used for indexing and search
When to use cloud instead
Turn Privacy Mode off when you want subscription cloud models (Claude, GPT, Gemini, Grok, and more) through your Elido plan. Local Semantic Index still keeps embeddings on-device either way.