Welcome to Faraday-OSS
Faraday is an AI scientist agent for multi-step life science research. Faraday-OSS is the open-source runtime: run it locally, connect your own tools and data, and keep a reproducible record of each run.
It is most useful when a task needs more than a single answer: inspect context, use tools, run code, generate files, and preserve the full workflow for review or reuse.
What you can do with Faraday
Section titled “What you can do with Faraday”- Run multi-step research tasks across literature, data, and code
- Save trajectories and artifacts for auditability, follow-ups, and evaluation
- Extend the agent with custom tools and workflows
- Benchmark agent behavior with a Harbor-ready integration
Start in 3 steps
Section titled “Start in 3 steps”- Install Faraday locally.
- Quickstart to run your first task.
- Reference Go deeper with configuring Faraday for specific setups and environments.
Why use Faraday
Section titled “Why use Faraday”- More than chat: the agent can reason, search, write code, and use tools over many steps
- Reproducible by default: each run leaves behind outputs, metadata, and a trajectory
- Flexible deployment: run locally or self-hosted when data control matters
- Built for extension: add domain-specific tools instead of overloading the prompt
Key Faraday use cases
Section titled “Key Faraday use cases”- Early-stage research triage such as literature review, competitive scans, and compound property screening
- Repeated analyses where you want comparable runs, batch execution, and saved artifacts
- Internal workflows that depend on proprietary files, custom tools, or team-specific heuristics
- Agent evaluation and process-data collection for training, benchmarking, or debugging
Credits
Section titled “Credits”Faraday-OSS is developed by Ascent Bio and is open-source under the MIT license.