Post-Training on Office Work Improves Software Engineering: A Behavioral Account of Cross-Domain Transfer
2026-08-03 • Artificial Intelligence
Artificial IntelligenceSoftware Engineering
AI summaryⓘ
The authors studied how training an AI model on long and complex tasks helps it get better at managing goals and steps over time. They focused on four key skills: picking goals, building relevant information, staying true to bigger objectives, and checking if tasks are done. After extra training, their model improved not just on similar office tasks but also on software engineering problems it hadn't seen before. This suggests that learning to handle long tasks helps the model organize and use knowledge more effectively across different types of work.
goal-directed executionlong-horizon taskspost-trainingQwen3.5-122B-A10BLHMTASWE-Bench Protrajectory analysisinformation gatheringtask verificationnested goals
Authors
Logan Ritchie, Sushant Mehta, Liudas Panavas, Edwin Chen
Abstract
Long-horizon tasks require agents to maintain coherent state and goals across nested and branching work. We call this capability goal-directed execution (GDE): the repeated application of four behaviors, namely selecting goals, constructing task-relevant state, maintaining fidelity to higher-level objectives, and verifying completion against the environment. We hypothesize that long-horizon post-training strengthens these behaviors across domains. We test this by post-training Qwen3.5-122B-A10B on 363 Long-Horizon Multi-Tool Agent (LHMTA) tasks drawn from office workflows. The collection contained no software-engineering tasks, yet the model's pass@1 improved by 5.8 points on SWE-Bench Pro. Matched trajectory analysis shows gains in all four GDE behaviors in both office workflows and software repositories. Aggregate SWE-Bench Pro statistics showed related changes in information gathering, implementation, and verification. Together, the results support a behavioral interpretation in which long-horizon post-training changed how the model organized and applied knowledge across tasks, with effects extending beyond the training domain.