DOC AI provider setup

Configure an AI provider.Choose the guide for your environment.

Use the documentation menu or choose a provider below. Each supported path—including Microsoft 365 Copilot—has its own complete setup guide.

01 Start here

Choose a provider.

Choose one category. Palshell keeps hosted providers, local inference, custom endpoints, and Microsoft’s organization-managed Copilot flow as separate setup paths.

Hosted provider

OpenAI

Fixed OpenAI endpoint, API key, model discovery, and the Responses API.Usage-based · API key required

Hosted provider

Anthropic

Fixed Anthropic endpoint, API key, Claude model discovery, and Messages API.Usage-based · API key required

Local inference

Ollama

A model running on this computer through the IPv4 loopback service.$0 local API · no key

Your endpoint

OpenAI-compatible

LM Studio, vLLM, LocalAI, or another compatible local or HTTPS server.Cost and key depend on server

Microsoft cloud

Microsoft 365 Copilot

Full Entra registration, delegated Graph permissions, administrator consent, and device-code sign-in.$0 extra with paid Copilot license

02 Quick reference

Open only the overview you need.

The full setup and troubleshooting details stay inside each provider category.

ONE RULE ACROSS EVERY CATEGORY

Configure providers in the desktop app—not on the website.

Provider secrets and Microsoft OAuth tokens stay in macOS Keychain or Windows Credential Manager. Palshell retrieves them only for the immediate provider operation, keeps AI context bounded and redacted, and still requires explicit approval before any proposed terminal command can run.

Read the security model

Back / all documentationBrowse every Palshell setup guide.