AliyunConsoleAgent: Training Web Agents in Real-World Cloud Environments via Distillation and Reinforcement Learning
2026-06-08 • Artificial Intelligence
Artificial Intelligence
AI summaryⓘ
The authors introduce AliyunConsoleAgent, a system designed to automatically check if cloud platform user interfaces match their documentation, a task that is very hard to do manually due to constant updates. They train their model using a special two-step process combining supervised learning and reinforcement learning with a new optimization method called GRPO. To train effectively, they build a stable testing setup with automated environment control and a rule-based system to fairly judge outcomes using backend logs. Their agent performs nearly as well as leading proprietary models but at a much lower cost, showing improved success in completing cloud console tasks.
cloud consoledocumentation verificationsupervised fine-tuningreinforcement learningGroup Relative Policy OptimizationTerraformreward modelbackend audit logslarge language modelsautomated testing
Authors
Bojie Rong, Zheyu Shen, Qiaoping Wang, Pengfei Kang, Yang Xu, Yawen Wei, Hanyu Wu, Zhi Zhao, Leihao Pei, Linquan Jiang
Abstract
We present AliyunConsoleAgent, a web agent framework for automated documentation verification in real-world cloud consoles. Major cloud platforms encompass hundreds of products with rapid feature iteration, causing console UIs to frequently diverge from their corresponding documentation. Verifying that documented procedures accurately reflect the current console and can be executed end-to-end demands an estimated 4 million recurring inspections annually, yet manual coverage remains below 1%. While agent systems built on frontier proprietary models achieve high success rates, their prohibitive cost and data privacy constraints preclude large-scale deployment. We propose a two-stage training paradigm: supervised fine-tuning (SFT) on distilled frontier-model trajectories, followed by reinforcement learning using Group Relative Policy Optimization (GRPO) and a dual-channel outcome reward model in real cloud environments. To support large-scale RL training, we construct a high-determinism rollout system featuring Terraform-based resource pre-provisioning and LLM-driven on-demand provisioning, which effectively isolates environment noise from the training signal. We further introduce a rule-based reward evaluation protocol grounded in backend audit logs, providing objective, reward-hacking-resistant outcome judgment. Our model evolves from mechanical instruction following to autonomous decision-making with cloud console and product-specific understanding. Experiments on a challenging 278-task benchmark where the best frontier model achieves only 65.34% demonstrate that AliyunConsoleAgent-32B achieves a 63.52% mean success rate -- a 20.24 percentage-point improvement over the base model, narrowing the gap to the best frontier proprietary model to 1.82 pp (bootstrap 95% CI [-1.27, 7.39]) -- at 92% lower inference cost.