OutSystems launches Agent Workbench to accelerate AI in business
OutSystems has introduced Agent Workbench, a platform designed to help enterprises develop and manage custom artificial intelligence agents across business operations.
The company has launched an Early Access Programme for Agent Workbench, intended to simplify the integration of AI agents into existing workflows and data environments with an emphasis on security and control. The tool aims to allow organisations to design, orchestrate, and monitor intelligent agents for various use cases without needing to overhaul their current systems.
Woodson Martin, Chief Executive Officer of OutSystems, described the need for Agent Workbench, noting the challenges faced by businesses seeking to implement AI technologies within enterprise settings.
"Organisations are excited by the promise of AI and agentic systems, but are struggling with endless pilots and ungoverned tool sprawl while the business impact stalls. Legacy systems, siloed data, fragmented AI tools, and complex AI development cycles are slowing progress," said Woodson Martin. "We built Agent Workbench to make it possible to unlock custom agents as a true business enabler - not an experiment."
Agent Workbench is delivered on the OutSystems platform and is positioned as a tool for enterprises to accelerate AI adoption and enable the integration of intelligent agents into day-to-day business operations. The platform is designed to transform existing business applications into agentic systems that can reason, plan, and act autonomously.
Features and integration
The OutSystems Agent Workbench provides several capabilities to help organisations deploy AI-powered agents across their departments, applications, and data environments. It allows IT teams to embed agentic AI into operations without disrupting current workflows or requiring significant architectural changes.
Organisations using Agent Workbench can build and scale AI agents throughout the enterprise, providing functionalities such as real-time goal interpretation, option evaluation, and decision-making. The platform offers a centralised approach to manage these capabilities, aiming to reduce the spread of disparate tools.
Agent Workbench supports integration with customised AI models or third-party providers, including Azure OpenAI and AWS Bedrock. This enables companies to centralise AI and data access while managing costs and supporting multi-vendor strategies.
Unified data and workflow orchestration
The platform grounds AI agents with a unified data fabric, allowing connections to a variety of data sources. These include existing OutSystems 11 data, relational databases, data lakes, knowledge retrieval systems such as Kendra and Azure AI Search, and agent memory of previous interactions. This is intended to ensure AI agents provide contextually relevant and accurate responses across workflows.
Orchestration of multi-agent workflows is also possible, where agents adapt their actions using real-time understanding of enterprise systems. Tasks can be handled collaboratively by agents in parallel, sequential, or hierarchical arrangements and allow for escalation and human intervention when required.
Governance and monitoring
Enterprise-wide monitoring of agent performance is supported through features like real-time logging, error tracing, and in-built guardrails for reliability and transparency. Organisations gain visibility into the operations of AI agents, which can help in auditing and troubleshooting agent behaviour, and in preventing erroneous outputs.
The OutSystems Agent Workbench Early Access Programme is open to organisations interested in participating before general availability later in the year.
