Project Jan 2025 1 min read

CellAgent

Live system page for CellAgent, highlighting the browser-based workflow for natural-language single-cell analysis and direct paths to the system and paper.

Related publication: CellAgent: LLM-Driven Multi-Agent Framework for Natural Language-Based Single-Cell Analysis · The Fourteenth International Conference on Learning Representations (ICLR 2026 Poster)

LLM Agent CellAgent

CellAgent is presented here as a usable research system rather than only a paper artifact. The live interface shows how researchers can move from a natural-language request to a structured single-cell analysis workflow in the browser.

Compared with the publication page, this detail view emphasizes the product-facing layer: how planning, tool orchestration, and domain-specific analysis are surfaced in an accessible workflow for researchers who do not want to stitch together custom scripts by hand.

If you want the method, evaluation, and technical framing, start with the paper. If you want to understand the interaction flow and deployment experience, start with the live system.

Jinyi Liu
Authors
Ph.D. Candidate Reinforcement Learning and LLM Systems