Husearch — AI that actually reads your sources
Ask a research question. Four specialist agents pull real data, trace
theoretical lineages, dig up historical precedents, and poke holes in the
argument — all at once, with citation verification built in. A multi-agent AI
research workspace for scholars writing books, papers, and essays in the
humanities and social sciences.
Start a project · Sign in
What it does
Submit a research question. Four specialist agents — Data, Theory, History,
and Critic — investigate in parallel, each with access to live academic and
primary-source APIs. A synthesizer agent merges their findings into a
structured research brief and annotated bibliography with verified citations.
How it works
- You ask a research question.
- Four parallel agents investigate different angles, calling external data and academic APIs as needed.
- A synthesizer resolves contradictions and writes a structured brief with citations.
- You annotate, write, and export — all in one workspace.
Features
- Parallel multi-agent research over academic papers, primary sources, and live economic data
- Annotated PDF reader with source criticism tools
- Rich-text manuscript editor with track changes and version history
- Real-time collaborative editing
- Knowledge graph and argument canvas
- Citation manager (CSL, BibTeX, RIS) with CrossRef verification
- Export to DOCX and LaTeX
- Per-project Research Focus configuration to steer agents to your discipline
Data sources
Academic: Semantic Scholar, OpenAlex, CrossRef, arXiv, CORE, Unpaywall.
Economic: FRED, World Bank, IMF, OECD, Eurostat, BIS, DBnomics, UN Comtrade.
Primary sources: Internet Archive, DPLA, Europeana, Library of Congress.
For researchers
Built for academic researchers, graduate students, and scholars who'd rather
think than search. No technical knowledge required — the interface stays out
of your way.
Questions, answered
- How is this different from ChatGPT? Agents call live academic and data APIs, so findings trace to real papers and datasets, and a Critic agent hunts for counter-evidence.
- How are fabricated sources avoided? Every cited claim links to a real source record, verified against CrossRef; unresolved sources are surfaced, never invented.
- Can I import my Zotero library? Yes — Husearch reads CSL-JSON, BibTeX, and RIS.
- Who owns my work? You do. Husearch does not train models on your projects, drafts, or notes.
- Can I export to LaTeX or Word? Yes — chapters or full manuscripts to .docx, print-ready PDF, or LaTeX.
- Does it work offline? Yes — the workspace is a PWA; drafts and recent sessions stay available offline.