Alon Oved

Alon Oved is an AI Research Scientist & Tech Lead at IBM Research–Haifa, where he designs, builds, and evaluates enterprise‑ready AI agents that reason, plan, and act autonomously. Within the Business Automation group, he leads projects bridging foundational research and real-world deployment in enterprise environments.

He co-created IBM CUGA (website: http://cuga.dev), a generalist agent architecture that supports both web browsing and API/tool integration for enterprise automation. CUGA holds 1st place on both the WebArena and AppWorld leaderboards, setting a new standard for robust, scalable agentic systems. Prior to that, Alon contributed to key innovations in AI-driven business process optimization, including the SNAP (https://ojs.aaai.org/index.php/AAAI/article/view/35153) framework for semantic next-activity prediction and ST-WebAgentBench (https://arxiv.org/abs/2410.06703), a benchmark suite for evaluating the safety and trustworthiness of LLM-based web agents.

Alon holds an M.Sc. in Computer Science (Specialized in AI and ML) from Imperial College London and a B.Sc. in Mathematics, Statistics, and OR from Tel Aviv University. His Master’s thesis led to the publication of EP-PINNs: Cardiac Electrophysiology Characterisation Using Physics-Informed Neural Networks, which introduced a novel AI method for non-invasively modeling cardiac electrical activity. The work enables personalized characterization of arrhythmias, such as atrial fibrillation, from sparse clinical data—supporting future innovations in diagnosis and treatment planning. (Read here: (https://www.frontiersin.org/journals/cardiovascular-medicine/articles/10.3389/fcvm.2021.768419/full) )