How can I use AI coding tools on private or sensitive repositories?

For private repos, your best options are: (1) use an enterprise plan with a DPA that prohibits training on your code, (2) use Tabnine or Continue.dev with a local model that never sends code to external servers, or (3) run a self-hosted LLM via Ollama. Many teams use a tiered approach — local models for sensitive internal code, cloud models for public-facing open source work.