Global Optimization and Inference-Time Region Grafting for Agentic Workflows
2026-08-03 • Computation and Language
Computation and Language
AI summaryⓘ
The authors present GRAFT, a method that improves workflows by changing only certain parts during execution instead of redesigning the whole process. GRAFT uses signals from how well tasks are doing without needing extra training or labels. This lets it adapt to different inputs efficiently and better than previous methods like MaAS. Their approach works on various tasks like math problems, coding, and complex question answering, showing that workflows can be flexible and improve over time with better tools.
workflow optimizationinference-time adaptationexecution-quality signalstask-specific workflowsagentic workflowslocal region replacementcode generationmulti-hop question answeringlabel-free evaluation
Authors
Donghyeok Koh, Gyuwan Kim, Jinyeong Bak, Seung-Hoon Na, Tao Yang, Haneol Jang, Cheoneum Park
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.