How to Use AI Research Agents?
Open AI Chat and pick an agent from the list above the chat input, or click an
agent in the left sidebar. Each run deducts credits based on your plan, and
results are saved to your chat history, library, and documents for later
reference.
How do you launch an AI research agent?
Open AI Chat and select the agent you want from the agent list shown above the chat input. You can also click an agent directly from the left sidebar. The agent starts right away and guides you through its workflow step by step. For the full list of available agents and what each one does, see What are AI Research Agents.
Can you stop an agent mid-run?
Yes. Click the stop button in AI Chat to cancel a running agent at any time. The agent stops immediately, and partial results from the stopped run are discarded — they will not appear in your chat history. Credits are deducted based on the actual computation used up to the stopping point, not the full expected cost.
How are AI research agents billed?
Agents are billed with credits that vary by plan. Each agent run deducts credits based on its computation cost. Some tools (paper search, peer review, proofreader) deduct credits based on actual LLM usage, while others deduct a fixed amount per run. See the pricing page for current plan allowances and details.
Where do you see results from past agent runs?
All results are saved in your chat history and stay accessible from the left sidebar. Papers collected by search agents are added to your library, and drafts generated by writing agents are stored in documents. You can revisit any past thread to continue or reference it in a new run.