Create Your Own Employee: Inside OutSystems Agent Workbench

Share this article
Share this article
Prioritise Us on Google
With Agent Workbench, companies will be able to customise agentic AI, tailoring agents to their own requirements
OutSystems has unveiled its ‘Agent Workbench’ platform which allows enterprises to build custom AI agents to streamlines operations and boost efficiency

Have you ever wished you had a clone that you could send into work when you need to see to something particularly laborious? That, in short, is the promise of agentic AI.

With this form of AI, users are able to double their capacity in an instant.

Gen AI has already been widely adopted across the working world, with companies using it to write emails, create eye-catching imagery or sort through huge data sets. Agentic AI is different, though. Agents have... well, agency.

Agentic AI can work completely independently of human oversight, meaning workers don’t need to feed agents prompts to dictate their actions — you just let them do their thing.

Of course, it is all a little more complicated and nuanced than this explanation might suggest. Thus far, some enterprises have struggled to create and deploy AI agents that are suited to their unique needs.

The applications for AI agents are seemingly limitless

To address this, OutSystems, a low-code enterprise IT platform, has announced the launch of its Agent Workbench system, designed to be an easy way of creating and customising agentic AI.

The Boston-based company positions the offering as a solution to what it sees as widespread struggles with AI implementation across sectors.

“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,” explains Woodson Martin, CEO of OutSystems.

“Legacy systems, siloed data, fragmented AI tools, and complex AI development cycles are slowing progress.

“We built Agent Workbench to make it possible to unlock custom agents as a true business enabler — not an experiment,” he says.

Woodson Martin, CEO of OutSystems (right) and Paulo Rosado, former CEO of Outsystems (left) | Credit: OutSystems

Enterprise-grade agent development

Agent Workbench allows organisations to transform existing business applications, workflows and tools into what OutSystems describes as intelligent, agentic systems capable of reasoning, planning and taking action.

The platform integrates with existing IT infrastructure without requiring wholesale system changes, potentially reducing implementation friction for enterprise customers.

IT teams can deploy agentic AI capabilities across enterprise operations whilst maintaining current workflows and core system architectures.

Agent Workbench supports integration with both custom AI models and third-party providers including Azure OpenAI and AWS Bedrock, giving organisations flexibility in their AI vendor relationships.

Youtube Placeholder

Data integration and workflow orchestration

The platform connects to diverse enterprise data sources, including OutSystems 11 data, relational databases, data lakes and knowledge retrieval systems such as Amazon Kendra and Azure AI Search.

This unified data fabric approach aims to provide AI agents with comprehensive context for more accurate responses across different business workflows.

Agent Workbench enables multi-agent workflows where different AI agents can collaborate on complex tasks, working in parallel, sequential or hierarchical configurations.

The system includes dynamic process flow adjustment capabilities, allowing agents to adapt their operations based on real-time understanding of enterprise systems.

Human intervention mechanisms are built into the workflow orchestration, providing escalation pathways when agent capabilities reach their limits.

“Agent Workbench is built for pragmatic leaders who are ready to put AI to real work and move beyond AI pilots to production, delivering business value,” says Ben Sekhon, VP for the UK & Ireland at OutSystems.

Ben Sekhon, VP for the UK & Ireland at OutSystems

Governance and monitoring measures

OutSystems has incorporated enterprise-grade monitoring capabilities into Agent Workbench, including real-time logging, error tracing and built-in guardrails.

These features are designed to provide transparency in AI decision-making processes and enable organisations to audit agent behaviour.

The platform includes mechanisms to prevent AI hallucinations and maintain ‘explainability’ (the ability to understand how an AI model arrived at its decisions) in agent operations, addressing common enterprise concerns about AI reliability.

Gonçalo Borrêga, VP for Product, AI & App Development at OutSystems

Market positioning and availability

It is a work-in-progress, but OutSystems are reaching out to businesses that want to try something new.

“We’re inviting forward-thinking teams to join our Early Adopters Program,” says Gonçalo Borrêga, VP for Product, AI & App Development at OutSystems.

“If you’re ready to be hands-on with AI agents, help us shape the experience, and push the boundaries of what companies around the globe can create with Agentic AI, we want to work with you.”