HyEvo: Self-Evolving Hybrid Agentic Workflows for Efficient Reasoning

2026-03-20

AI Summary

Existing approaches to automatic agent workflow generation rely on homogeneous LLM pipelines and predefined operator libraries, resulting in high inference costs and substantial latency. This work proposes HyEvo, a novel framework that introduces a heterogeneous hybrid workflow architecture, decoupling deterministic computation from LLMs and synergistically combining LLM-driven semantic reasoning with rule-based execution in code nodes. HyEvo further incorporates an LLM-guided multi-island evolutionary algorithm, which iteratively refines both workflow structure and node logic through a reflection-generation mechanism enriched with execution feedback. Evaluated across multiple reasoning and programming benchmarks, HyEvo significantly outperforms state-of-the-art methods, achieving up to a 19× reduction in inference cost and a 16× decrease in execution latency.

Technology Category

• Humans and AI: Human-Aware Planning and Behavior Prediction
• Machine Learning: Evolutionary Learning
• Search and Optimization: Evolutionary Computation

Application Category

• Economics, Online Markets and Human Computation: Architectures and workflows that use LLMs for crowd work
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs

Abstract

Although agentic workflows have demonstrated strong potential for solving complex tasks, existing automated generation methods remain inefficient and underperform, as they rely on predefined operator libraries and homogeneous LLM-only workflows in which all task-level computation is performed through probabilistic inference. To address these limitations, we propose HyEvo, an automated workflow-generation framework that leverages heterogeneous atomic synthesis. HyEvo integrates probabilistic LLM nodes for semantic reasoning with deterministic code nodes for rule-based execution, offloading predictable operations from LLM inference and reducing inference cost and execution latency. To efficiently navigate the hybrid search space, HyEvo employs an LLM-driven multi-island evolutionary strategy with a reflect-then-generate mechanism, iteratively refining both workflow topology and node logic via execution feedback. Comprehensive experiments show that HyEvo consistently outperforms existing methods across diverse reasoning and coding benchmarks, while reducing inference cost and execution latency by up to 19$\times$ and 16$\times$, respectively, compared to the state-of-the-art open-source baseline.

Problem

• agentic workflows
• automated workflow generation
• heterogeneous reasoning
• LLM inference cost
• execution latency

Innovation

• hybrid agentic workflows
• heterogeneous atomic synthesis
• LLM-driven evolutionary search
• deterministic code nodes
• reflect-then-generate mechanism