Institution-friendly workflows (local and approved chat)

[kenflab]

We provide two end-to-end, institution-policy–friendly workflows that keep raw matrices and cell-level metadata local and avoid external LLM API calls from the tool unless explicitly permitted:

1. Fully local LLM (Ollama) (no external transmission): curate features and optionally annotate clusters using a local LLM backend(./examples/local/local_quickstart_ollama.ipynb).

2. Local feature distillation → institution-approved chat annotation (no tool-side API calls): curate features locally, export a curated cluster→genes table, then annotate via an approved chat interface (e.g., Copilot “Work”) by uploading CSV/Excel or pasting markers(./examples/local/local_quickstart_approved_ai_workflow.ipynb).