BUDDY AI documentation
Everything you need to get the most out of BUDDY AI — from your first message to agents, research, files and the developer platform.
Create an account, open a chat, and type your first message. Pick a mode — FAST, THINK, CODE, RESEARCH, CREATIVE — or leave it on AUTO and BUDDY will choose. Upload a file to ask about it, or give the Super Agent a goal to complete end to end.
Chat streams responses in real time with markdown, code blocks, tables and math. Stop or regenerate any response. Substantial code or documents open in the artifact panel where you can refine and download them. Organize conversations with pin, archive, rename and search.
Give BUDDY a goal and the Super Agent builds a task graph, runs the steps, and synthesizes a result. Independent tasks run in parallel; high-impact steps pause at an approval gate where you can approve, modify or reject. You can cancel a run at any time and review every artifact and activity entry.
Research mode retrieves information from multiple sources and answers with inline citations you can verify. Combine web research with your uploaded documents for answers grounded in both public and private context.
Upload PDFs, Word docs, spreadsheets, presentations, CSVs, text, JSON, code and images. Summarize, extract tables and data, generate reports, or ask questions grounded in the file. Images are understood with vision.
Studio is a focused workspace for image generation. Describe what you want, generate it, and keep results in your personal library. Images can also be generated inline in chat and opened in Studio.
CODE mode is tuned for software: generating, explaining, refactoring and debugging across many languages. Larger outputs open in a code artifact panel you can copy or download, and projects keep context consistent.
Create, list and revoke API keys from your account and explore the API reference. Programmatic gateway access is being rolled out; where a capability is not yet live it is shown honestly as configuration-required.
BUDDY Connect for external services (such as Google, GitHub and Slack) is on the roadmap and requires configuration before use. Until an integration is configured, it is shown as configuration-required rather than simulated.
Your data is isolated to your account. Provider secrets are never exposed to the client, AI activity and security events are audit-logged, and long-term memory is fully under your control. A unified safety policy applies across chat, studio and agents.