De-Linearizing Agent Traces: Bayesian Inference of Latent Partial Orders for Efficient Execution

2026-02-02

AI Summary

This work addresses the inefficiency arising from the sequential execution of AI agents, which obscures inherent concurrency and leads to redundant reasoning. To tackle this, the authors propose the BPOP framework, which introduces Bayesian partial order learning for the first time to de-linearize agent execution traces. By modeling observed trajectories as random linear extensions of an underlying latent partial order dependency graph and employing a novel frontier-aware Softmax likelihood function, BPOP enables efficient MCMC inference while circumventing the #P-hard marginalization problem. The resulting partial order structure facilitates compiled concurrent execution, significantly reducing both token consumption and runtime in LLM calls. Empirical evaluation on the Cloud-IaC-6 and WFCommons datasets demonstrates that BPOP accurately recovers ground-truth dependency structures, achieving substantial gains in execution efficiency.

Technology Category

• Reasoning under Uncertainty: Sequential Decision Making
• Humans and AI: Human-Aware Planning and Behavior Prediction
• Search and Optimization: Sampling/Simulation-based Search

Application Category

• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• Economics, Online Markets and Human Computation: Cost models of using LLMs in production systems
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs

Abstract

AI agents increasingly execute procedural workflows as sequential action traces, which obscures latent concurrency and induces repeated step-by-step reasoning. We introduce BPOP, a Bayesianframework that infers a latent dependency partial order from noisy linearized traces. BPOP models traces as stochastic linear extensions of an underlying graph and performs efficient MCMC inference via a tractable frontier-softmax likelihood that avoids #P-hard marginalization over linear extensions. We evaluate on our open-sourced Cloud-IaC-6, a suite of cloud provisioning tasks with heterogeneous LLM-generated traces, and WFCommons scientific workflows. BPOP recover dependency structure more accurately than trace-only and process-mining baselines, and the inferred graphs support a compiled executor that prunes irrelevant context, yielding substantial reductions in token usage and execution time.

Problem

• agent traces
• latent concurrency
• partial order
• workflow execution
• linearization

Innovation

• Bayesian inference
• partial order
• linear extensions
• frontier-softmax
• workflow execution