AgentCo-op: Retrieval-Based Synthesis of Interoperable Multi-Agent Workflows

2026-05-19 · Source: Artificial Intelligence · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Life Sciences & Biology · Depth: Expert, quick

Summary

AgentCo-op is a novel retrieval-based synthesis framework designed to simplify the creation of multi-agent workflows, particularly in complex, open-ended scientific domains lacking curated training data or standardized interfaces. It achieves this by composing reusable skills, tools, and external agents into executable workflows using typed artifact handoffs, and employs bounded self-guided local repair when execution failures occur. In two open-world genomics case studies, AgentCo-op successfully integrated independently developed scientific agents for tasks like spatial transcriptomics and gene-set interpretation, and built parallel workflows for cross-modality marker analysis on single-cell multiome data. The framework also demonstrated the ability to import and refine existing workflows. On six coding, math, and question-answering benchmarks, AgentCo-op secured the best result on four and the best average score under a unified backbone setting, while consistently reducing per-task cost compared to multi-agent baselines.

Key takeaway

For research scientists or ML engineers developing multi-agent systems for open-ended scientific tasks, AgentCo-op offers a robust approach to workflow synthesis. You should consider integrating retrieval-based methods and typed artifact handoffs to build auditable, interoperable workflows from existing agents and tools. This can significantly reduce development costs and improve reliability, as demonstrated by its performance on genomics and benchmark tasks, allowing you to focus on novel agent development rather than interface design.

Key insights

Retrieval-based synthesis enables robust multi-agent workflow composition and repair in open-ended scientific and general domains.

Principles

• Retrieval-based synthesis extends agentic workflow design to open-world scenarios.
• Typed artifact handoffs ensure interoperability between diverse components.
• Bounded local repair addresses execution failures in complex workflows.

Method

AgentCo-op composes reusable skills, tools, and external agents into workflows via typed artifact handoffs, applying bounded self-guided local repair upon execution failure. It can also refine prior searched workflows.

In practice

• Coordinate specialized agents for spatial transcriptomics and gene-set interpretation.
• Build parallel workflows for cross-modality marker analysis on single-cell multiome data.
• Improve existing searched workflows by grounding nodes and applying local repair.

Topics

• Multi-Agent Workflows
• Retrieval-Based Synthesis
• Scientific Discovery
• Genomics
• Workflow Automation
• Local Repair

Best for: AI Engineer, AI Scientist, Machine Learning Engineer, Research Scientist

Related on AIssential

• From Specification to Execution: AI Assisted Scientific Workflow Management — AI-assisted workflow management streamlines design, execution, and debugging through structured specification and LLM agents.
• A Coding Guide to Design and Orchestrate Advanced ReAct-Based Multi-Agent Workflows with AgentScope and OpenAI — Orchestrate specialized ReAct agents with AgentScope for robust, collaborative multi-agent workflows.
• Announcing Lakebase Search: agent-native retrieval built into Lakebase Postgres — Agent-native search requires instant, real-time retrieval on a single, cost-effective, tiered storage backend.
• An End-to-End Guide to Beautifying Your Open-Source Repo with Agentic AI — OSA uses LLM agents to holistically automate open-source repository documentation, CI/CD, and structural improvements.
• The Agent Stack - Part 4: Runtimes, Workflows, and Durable Execution — The runtime layer is crucial for managing complex, long-running agent operations by ensuring progress and recoverability.
• not much happened today — AI agent tooling, infrastructure, and robotics are rapidly advancing, while platform stability and search access pose new challenges.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Artificial Intelligence.