Decentralized agents accelerate diverse idea generation in scientific discovery

Inventa: Decentralized Self-Improving Agents for Autonomous Scientific Discovery

Machine LearningArtificial Intelligence

Summary

Generating new scientific ideas can be tough because groups working together may start copying each other or get stuck doing the same things repeatedly. The authors created Inventa, a system where many independent agents work on creating and testing ideas separately but share only verified facts. This helps keep ideas fresh and more reliable. Their experiments show Inventa produces more new ideas and solves scientific problems better than previous methods, using fewer resources.

What this means in practice

  • For software developers: Develop AI software that autonomously generates and refines creative solutions in technical domains with verified sharing to avoid redundant work.
  • For product innovation teams: Use decentralized agent frameworks to explore and validate novel product concepts faster by automating idea evolution and collaborative verification.

Authors

Prithwish Jana, Nikos Kanakaris, Purak Jain, Anoop Deoras, Sahika Genc

Abstract

Large language model (LLM) agents can generate research ideas and improve scientific discovery through iterative experimentation. However, increasing the number of agents or iterations does not automatically translate to producing more diverse or feasible ideas. Agents may converge on similar candidates, propagate unsupported claims through shared memory components and repeatedly apply search strategies that no longer yield any improvements. We introduce Inventa, a framework for automated scientific discovery that employs autonomous research units and verified shared knowledge. Each unit creates a lineage of candidate ideas using a mutator agent. A local orchestrator agent periodically adapts the mutator's harness. When progress stalls, the orchestrator agent replaces its model or restarts the idea evolution process. Verifiers evaluate each candidate idea and record verified improvements, useful components and constraint violations in a shared knowledge, the only channel between units. We evaluate Inventa on five benchmarks spanning the research cycle, covering retrieval (ScholarCatalyst), ideation (IdeaBench), proposal writing (IdeaScientist), and scientific law discovery (EurekaBench, SciLaws-Bench). It leads all five, achieving 73.6% on EurekaBench versus 56.2% for CORAL. On IdeaBench, it discovers 26.0 novel ideas per paper, compared with 18.6 for ResearchAgent, with 82% fewer model calls.