GLOW: Graph-Language Co-Reasoning for Agentic Workflow Performance Prediction
This work addresses the scalability issue in automating AW generation for complex tasks, offering a more efficient prediction method, though it appears incremental as it builds on existing GNN and LLM techniques.
The paper tackles the problem of predicting the performance of Agentic Workflows (AWs) by addressing the limitations of existing methods that fail to capture both topological dependencies and semantic logic, proposing GLOW, a framework that combines GNNs and LLMs, which outperforms state-of-the-art baselines on FLORA-Bench in accuracy and ranking utility.
Agentic Workflows (AWs) have emerged as a promising paradigm for solving complex tasks. However, the scalability of automating their generation is severely constrained by the high cost and latency of execution-based evaluation. Existing AW performance prediction methods act as surrogates but fail to simultaneously capture the intricate topological dependencies and the deep semantic logic embedded in AWs. To address this limitation, we propose GLOW, a unified framework for AW performance prediction that combines the graph-structure modeling capabilities of GNNs with the reasoning power of LLMs. Specifically, we introduce a graph-oriented LLM, instruction-tuned on graph tasks, to extract topologically aware semantic features, which are fused with GNN-encoded structural representations. A contrastive alignment strategy further refines the latent space to distinguish high-quality AWs. Extensive experiments on FLORA-Bench show that GLOW outperforms state-of-the-art baselines in prediction accuracy and ranking utility.