Research Agent
The research agent runs multi-step scholarly workflows from your project context: search, source triage, reading, synthesis, and follow-up questions.
What the agent is for
The research agent is designed for questions that need more than a single search box or one chat response. It can work from your papers, notes, drafts, citations, and project memory, then decide which searches, sources, and checks are needed next.
Find relevant work
Read and triage sources
Explore in parallel
Leave an audit trail
How it differs from chat
Chat is best when you want help with the document in front of you. The research agent is best when the answer depends on searching, comparing sources, and keeping track of what was found. It still talks through the chat surface, but it runs a structured workflow behind the response.
- Persistent contextThe agent can use the project record instead of asking you to restate the same background each turn.
- Tool-aware workflowIt can search literature, read project files, inspect citations, and create report artifacts when the task calls for it.
- Source-backed outputAnswers are expected to point back to papers, passages, and decisions rather than only presenting polished prose.
Common workflows
- Research reportsAsk a question and get a structured literature overview with screening, extraction, and citations.
- Gap checksAsk what a reviewer would say is missing from a draft, methods section, or bibliography.
- Contradiction checksCompare a claim in your draft against evidence from your library and adjacent papers.
- Corpus updatesAsk what changed since your last review or digest, then decide what belongs in the project.
Privacy and control
The research agent runs inside the same project permissions as the rest of Academe. It can only use project content you have access to, and project context stays scoped to your account or team according to your workspace permissions.