Manage literature, track experiments, and generate structured research updates automatically — the entire scientific loop, accelerated.

§ 01 Built for researchers
Papers, advisor meetings, and writing deadlines pulling in every direction.
Literature, experiments, and progress notes in one place — with weekly advisor updates generated in a click.
Chasing status across students, projects, and grant deadlines every week.
Progress, risks, and next steps written up for group meeting — plus funding strategy and proposal drafts.
Experiment runs scattered across notebooks, spreadsheets, and memory.
Every run logged with outcomes, blockers, and next steps — all your work from one dashboard.
Log in or create an account to start managing your research workspaces.
§ 02 The agent that does the work
Flow is in every workspace. It reads what you have imported, runs the tools it needs, and shows you which sources each answer came from.

Semantic search across your papers, notes, manuscripts and artifacts — and arXiv when the answer is not in the workspace yet.
Citation-backed synthesis, frontier and gap mapping, red-team critique of a draft, and a citation review that finds uncited claims.
Executes Python in an isolated sandbox — pandas, numpy, matplotlib — and hands back the output and the plots it drew.
Drafts notes, tasks, experiment logs and manuscript sections — each one held as a proposal until you approve it.
§ 03What's inside
Ask in plain language. It searches, analyses and drafts — and waits for approval before it writes.
Import by DOI or URL. AI cards: summary, methods, findings, limitations.
Log runs, outcomes, blockers, next steps — artifacts on every run.
Citation-backed reviews, weekly updates, structured plans — one click.
Kanban with due-dates, overdue alerts, bulk moves, inline editing.
Papers, tasks, notes, artifacts — across every workspace, instantly.
Free, Researcher $29/mo, Team $99/mo. Stripe-powered billing.
Clone structures as templates. Export Markdown, JSON, or CSV.
JWT auth with rotating refresh, scoped API tokens, rate-limited AI endpoints.
§ 04 How a week actually runs
Sync a Zotero or Mendeley library, add papers by DOI, and attach the datasets, figures and notebooks they came from.
One view of what changed this week, what is overdue, and which experiments are still running.
A generated narrative of progress, open questions and priorities — the group-meeting update, drafted for you.
Overdue and due-soon tasks in priority order, so the next thing to do is never a decision.
§ 05 Start from a proven structure
§ 06 See it working




§ 07 Works with what you already use
Your references stay where they are. ResearcherFlow reads from them — it does not ask you to move your library.
§ 08 Questions people actually ask
No. Anything that writes — a note, a task, a manuscript section — is held as a pending action for you to approve or reject. The tool does not run until you say so, and every run keeps a trace of what it did and why.
Yes, and without asking us. Workspaces export as Markdown or JSON, papers, tasks, notes, questions, experiments and artifacts export as CSV, manuscripts as Markdown, and slide outlines as PPTX. You can also delete your account and its data outright.
Zotero and Mendeley libraries sync in, including collections and citations. GitHub repositories and Jupyter notebooks are indexed from links you add, so their content is searchable alongside your papers. Your library stays where it is.
Anthropic's Claude Sonnet, through OpenRouter. Usage and cost are metered per workspace so you can see what the AI actually spent, and the model is configurable if your institution requires a different one.
Three workspaces with 100 papers each, and the AI features on top — enough to run a real project, not a demo. Researcher and Team raise those limits to 1,000 and 10,000 papers per workspace.
Only people you invite, with per-member roles from viewer to owner. Nothing is public unless you publish it: sharing a manuscript or artifact is an explicit, owner-only toggle on that one item.
Move from noise to signal.