Documentation

By default DataPilot answers with its own models. If you'd rather use your own LLM account, add your API key and every Ask & Code Agent call runs through it instead — you keep full control of usage and billing on the provider's side.

Which providers

• Anthropic — a Claude API key (sk-ant-…).
• OpenAI — an OpenAI API key (sk-…).
• Custom / self-hosted — any OpenAI-compatible endpoint: you provide a Base URL and a Model name (e.g. a local vLLM or Ollama server).

How your key is used

• Encrypted at rest. Your key is stored encrypted (Fernet) and scoped to your account only — never shared with other users.
• Never logged. It is not written to logs, analytics, or error reports.
• Only ever sent to the provider you chose. An Anthropic key is only sent to Anthropic; an OpenAI key only to OpenAI; a custom key only to the Base URL you entered. It is never sent anywhere else.
• Used only for your LLM calls — the answers and code your account generates. It is not used for embeddings, indexing, or any background job.
• Removable anytime. Clear the key to instantly fall back to DataPilot's built-in models.

Add a key

1. In the app's top navigation, click the key button.
2. Pick your provider, paste your API key (and Base URL + Model for a custom endpoint).
3. Click Test key to confirm it works, then Save key.

Point DataPilot at a GitHub repository and ask questions about your code — grounded, with citations back to the exact file. On the Free plan you can connect one repo (ask-only); Pro and Premium connect unlimited repos.

• Public repos need no token. Private repos need a GitHub access token so DataPilot can read them.
• Ask-only: DataPilot reads and answers from your code — it does not push changes back to the repo.

Everything you upload is stored privately to your account and used only to answer your questions. Answers cite the exact page, row, or file they came from, so you can always check the source.

Built-in agent pages draw from a shared, curated library that is the same for everyone; your own uploads are never part of it unless you publish them.

Every plan can reach us through the in-app Report a bug and Feedback buttons — reports go straight to the team's inbox.

Premium SLA (in writing): Premium accounts get priority support — we respond to Premium support requests within 1 business day. If we miss that window in a given month, tell us and we'll credit 10% of that month's fee. Business days are Monday–Friday, US time.

This page is the written commitment the Premium plan's "Priority support & SLA" refers to.