Towards Enterprise-Ready Computer Using Generalist Agent
This work addresses the problem of building scalable and efficient agentic AI systems for enterprise environments, representing an incremental advancement in integrating existing techniques.
The paper tackles the development of an enterprise-ready Computer Using Generalist Agent (CUGA) system, achieving new state-of-the-art performance on WebArena and AppWorld benchmarks through iterative evaluation and refinement.
This paper presents our ongoing work toward developing an enterprise-ready Computer Using Generalist Agent (CUGA) system. Our research highlights the evolutionary nature of building agentic systems suitable for enterprise environments. By integrating state-of-the-art agentic AI techniques with a systematic approach to iterative evaluation, analysis, and refinement, we have achieved rapid and cost-effective performance gains, notably reaching a new state-of-the-art performance on the WebArena and AppWorld benchmarks. We detail our development roadmap, the methodology and tools that facilitated rapid learning from failures and continuous system refinement, and discuss key lessons learned and future challenges for enterprise adoption.