Global Optimization and Inference-Time Region Grafting for Agentic Workflows
Donghyeok Koh, Gyuwan Kim, Jinyeong Bak, Seung-Hoon Na, Tao Yang, Haneol Jang, Cheoneum Park
cs.CL
Submitted: 2026-08-03
Comments: 9 pages, 3 figures, 4 tables
License: http://creativecommons.org/licenses/by/4.0/
The gist: Recent advances in agentic workflow optimization automate workflow design through task-specific workflow search or input-conditioned architecture selection.
Terminology
Abstract
Recent advances in agentic workflow optimization automate workflow design through task-specific workflow search or input-conditioned architecture selection. However, they determine the workflow before execution and cannot adapt failed workflow regions using execution-time label-free quality signals. Naively enabling such inference-time adaptation through whole-workflow re-optimization would be computationally prohibitive. To tackle this challenge, we introduce GRAFT, which preserves a globally optimized workflow while locally replacing only selected regions for each input. Without parameter training, GRAFT evaluates region-level alternatives using label-free execution-quality signals and accepts only replacements that improve local quality while preserving workflow-level consistency, thereby enabling instance-wise adaptation without whole-workflow re-optimization. GRAFT applies without modification across a range of tasks spanning mathematical reasoning, code generation, and multi-hop and knowledge-intensive question answering. Under matched optimizer and executor settings, it improves over the strongest prior workflow-optimization method, MaAS, by 3.85 points on average. Replacing only the executor with a stronger model yields further gains without re-optimizing the global workflow. This suggests that an optimized workflow is not merely a static optimization artifact, but an adaptable execution policy that can evolve with inference-time feedback and stronger executors.
Sources
- Program Synthesis with Large Language Models
- Why Do Multi-Agent LLM Systems Fail?
- Evaluating Large Language Models Trained on Code
- AgentVerse: Facilitating Multi-Agent Collaboration and Exploring Emergent Behaviors
- Training Verifiers to Solve Math Word Problems
- MetaGPT: Meta Programming for A Multi-Agent Collaborative Framework
- Automated Design of Agentic Systems
- CAMEL: Communicative Agents for "Mind" Exploration of Large Language Model Society
- AgentSwift: Efficient LLM Agent Design via Value-guided Hierarchical Search
- LLM+P: Empowering Large Language Models with Optimal Planning Proficiency
- A Dynamic LLM-Powered Agent Network for Task-Oriented Agent Collaboration
- Large Language Models: A Survey
- Flow: Modularized Agentic Workflow Automation
- AgentSquare: Automatic LLM Agent Search in Modular Design Space
- DyFlow: Dynamic Workflow Framework for Agentic Reasoning
- ScoreFlow: Mastering LLM Agent Workflows via Score-based Preference Optimization
- AutoGen: Enabling Next-Gen LLM Applications via Multi-Agent Conversation
- The Rise and Potential of Large Language Model Based Agents: A Survey
- EvoFlow: Evolving Diverse Agentic Workflows On The Fly
- AFlow: Automating Agentic Workflow Generation
Related papers
- Exploring Solution Divergence and Its Effect on Large Language Model Problem Solving
- Ishigaki-IDS-Bench: A Benchmark for Generating Information Delivery Specification from BIM Information Requirements
- Subliminal Steering: Stronger Encoding of Hidden Signals
- MedStruct-S: A Benchmark for Key Discovery, Key-Conditioned QA and Semi-Structured Extraction from OCR Clinical Reports
- The End of Transformers? On Challenging Attention and the Rise of Sub-Quadratic Architectures
- Untangling the Mechanisms of Misleading Context in Medical Question Answering