Interested in using AI to accelerate your research?
Here are some of the places we love for doing that: AI co-scientists that help generate new work, AI reviewers that critique it, and autonomous labs that run the experiments. After you’ve generated your work with these systems, come back and publish it on Valency Hub.
31 systems
AI co-scientists 18 systems
Agents that read the literature, form hypotheses, run analyses and draft the write-up with you.
-
Harness that gives any LLM agent checkable tools for exact quantitative calculations.
-
Modular multi-agent research assistant that takes a project from idea to draft paper.
-
General-purpose biomedical agent that plans and runs analyses across hundreds of tools and databases.
-
Multi-agent system that automates hypothesis generation and experiment design for drug repurposing.
-
Gemini multi-agent system that generates, debates and ranks novel research hypotheses.
-
Autonomous AI scientist that runs 12-hour cycles of data analysis and literature review.
-
Agent that designs lab-ready therapeutic antibodies from a text prompt.
-
Chemistry agent wired to expert tools for synthesis planning and molecule design.
-
Chemist-X (opens in a new tab) CUHK · Zhejiang
Retrieval-augmented agent that recommends reaction conditions and runs them on a synthesis robot.
-
LLM-RDF (opens in a new tab) Zhejiang University
Six agents that guide synthesis development end to end, from literature search to scale-up.
-
Multi-agent platform for de novo protein design, backed by physics simulations.
-
Multi-agent research engine for cosmology data analysis; the backend behind Denario.
-
Agentic tree search that generates ideas, runs ML experiments and writes the paper.
-
Agent Laboratory (opens in a new tab) AMD · Johns Hopkins
LLM agents that turn your research idea into experiments, code and a report.
-
End-to-end autonomous research, from literature review and ideas to code and paper writing.
-
AgentRxiv (opens in a new tab) Johns Hopkins · ETH Zurich
Shared preprint server where agent labs publish and build on each other's results.
-
High-accuracy agent that answers questions from the scientific literature, with citations.
-
Library of hundreds of scientific tools that turns any LLM into a research agent.
AI reviewers 10 systems
Referee agents and pre-submission checks that critique a manuscript before anyone else sees it.
-
paperreview.ai (opens in a new tab) Stanford Agentic Reviewer
Free agentic reviewer that gives actionable, literature-grounded feedback on a paper draft.
-
OpenReviewer (opens in a new tab) L3S · Leibniz Univ. Hannover
Open 8B model fine-tuned on 79K ICLR and NeurIPS reviews to write critical reviews.
-
Reviewer2 (opens in a new tab) Cornell
Two-stage model that predicts what reviewers will focus on, then writes the review.
-
Multi-agent pre-submission check of citations, statistics, code and claims.
-
Referee that reviews passage by passage, flagging math errors, overclaims and vague methods.
-
Simulates peer review with reviewer, author and chair agents, to study reviewer bias.
-
Open models trained with RL to write ML papers and simulate their peer review.
-
Reasoning reviewer that searches the literature, simulates several reviewers and checks itself.
-
Several agents split a long paper between them to write specific review comments.
-
Deep pre-submission check for errors in math, consistency, citations and reasoning.
Autonomous labs 3 systems
Robotic and cloud labs that run the experiments themselves.
-
AI–XR co-scientist: agents and smart glasses that guide and check wet-lab work.
-
Remote cloud lab: write experiments as code and run them on 200+ instrument models, 24/7.
-
LLM agent that plans and executes chemistry experiments on lab robots.
Made something new with these?
Publish it on Valency Hub, and credit the systems that helped in its AI-use disclosure.
Know a system we’re missing? Tell us about it.
Valency isn’t affiliated with these projects, and a listing isn’t an endorsement. Descriptions are ours; each link opens the project’s own site in a new tab, and logos belong to their owners. Access and licence details were checked on 4 October 2026 and can change.