OutSystems announced the Early Access Program for OutSystems Agent Workbench which simplifies the transformation of existing business applications, workflows, and tools into intelligent, agentic systems that can reason, plan, and act. With Agent Workbench, organizations of any size can: Build and scale AI agents that work across the entire enterprise effortlessly and safely for real-time goal interpretation, option evaluation, and decision-making, while controlling tool sprawl through a single platform. Seamlessly integrate with custom AI models or leading third-party providers like Azure OpenAI and AWS Bedrock to centralize control over AI and data access, decrease cost, and enable multi-vendor utilization. Ground AI agents with a unified data fabric that connects to diverse enterprise data sources, such as existing OutSystems 11 data and actions, relational databases, data lakes, knowledge retrieval systems like Kendra and Azure AI Search, and even agent memory of past interactions, to ensure accurate and context-rich responses across workflows. Orchestrate multi-agent workflows where agents dynamically adjust process flows based on an understanding of all enterprise systems, with real-time context, reasoning, and decisions to tackle complex tasks—whether working in parallel, sequentially, or hierarchically. This enables collaborative task execution, escalation handling, and human intervention when necessary. Monitor agent performance enterprise-wide with real-time logging, error tracing, and built-in guardrails to ensure transparent, reliable decision-making. Gain full visibility into how AI agents operate at every step—making it easy to audit, troubleshoot behavior, and prevent hallucinations, while building trust through explainability and control.