AutoResearch vs Curie
AI-assisted research workflows compared
AutoResearch and Curie use Python agents to support scientific and machine-learning experiments, from planning through analysis. AutoResearch emphasizes traceable, resumable work and independent review, while Curie focuses on automating experiments with user-provided datasets or starter code and producing reports.
AutoResearch: Turn Research Ideas into Reviewable Evidence
AutoResearch is a Python agent workflow for developing AI and machine learning research ideas into experiments and reviewable evidence. It suits researchers who want a traceable, resumable process and can configure model services and experiment resources.
Curie: Automate Scientific Experiments with AI Agents
Curie is a Python framework that uses AI agents to plan, run, analyze, and report on scientific experiments. It suits researchers who want to automate exploratory or machine-learning work, but its inactive development and lack of recent releases warrant caution.
| AutoResearch | Curie | |
|---|---|---|
| Language | Python | Python |
| License | Apache-2.0 | Apache-2.0 |
| Stars | 2.9k | 370 |
| Forks | 256 | 32 |
| Last analyzed | Oct 4, 2026 | Oct 9, 2026 |
Key differences
- AutoResearch can discover research directions from papers, community discussions, and open-source trends; Curie centers on investigating a research question, including with a user's dataset or starter code.
- AutoResearch records plans, queues, logs, results, and decisions for resuming work, and supports pilot experiments and preservation of negative results; Curie emphasizes verification modules, experiment artifacts, and generated reports.
- AutoResearch's Idea Forge requires at least three distinct models for its multi-model stage; Curie describes model credentials and API keys without that stated requirement.
- AutoResearch's execution workflow requires Bun, Node.js, and an experiment environment manager such as Conda, and warns that its current execution entrypoint grants Claude Code broad tool permissions; Curie requires Docker for experiments.
- Both projects use Python and have an Apache-2.0 license. AutoResearch lists 2.9k stars and 256 forks, while Curie lists 370 stars and 32 forks.
Choose AutoResearch if you…
- need a durable record of experiment plans, logs, results, and decisions that supports resuming work.
- want to generate research directions from current research signals or a hypothesis, then challenge conclusions with independent reviews.
- want to run a pilot before scaling an experiment and retain failure causes or negative results.
Choose Curie if you…
- want an agent to investigate a focused question using your own dataset or existing starter code.
- need automated experiment reports and verification modules intended to support reproducible procedures.
- want to explore machine-learning approaches with Curie's AutoML feature or use it for system analysis.
Compare the numbers side by side, or add more projects
This comparison is generated with AI from the OSRepos analyses of both projects. Always check each project's repository and documentation before choosing.