Blab Tech
OutSystems Unveils Open Agentic Systems Platform at ONE Conference
At the 2026 ONE Conference in Amsterdam, OutSystems—an early‑adopter of low‑code development since 2001—unveiled a bold new suite of AI‑centric products. The company framed the launch as a pivot toward an “agentic systems platform” that can orchestrate AI agents across a company’s existing software stack without owning the underlying data.
The centerpiece is the OutSystems Agent Experience, a platform layer that exposes two open protocols: the Model Context Protocol (MCP) and the Agent‑to‑Agent (A2A) services. MCP, introduced by Anthropic in 2024, lets large language models (LLMs) pull in external data and tools, while A2A—led by Google—enables distinct AI agents to talk to one another. By weaving these standards together, OutSystems claims developers can build, publish, and extend applications with third‑party coding tools such as Claude Code, Codex, Cursor, Kiro, and AWS’s spec‑centric IDE.
The first MCP‑enabled service is already live on the OutSystems Developer Cloud (ODC), OutSystems’ cloud‑native platform. Support for the older, self‑managed OutSystems 11 version was also released in early access. CIO Tiago Azevedo explained that the move gives legacy customers “the ability to use Claude or Codex or whatever to evolve my applications in O11.”
Beyond the Agent Experience, the company introduced Agentic Enterprise Orchestration, a new layer built on its Enterprise Context Graph. The graph stores real‑time organizational data and serves as a foundation for agent reasoning. The orchestration service adds guardrails, semantic search, and Amazon Bedrock support, empowering enterprises to manage agent workflows at scale.
OutSystems also previewed a modernization service that leverages AWS Transform and Kiro to migrate legacy systems—such as COBOL and Lotus Notes—onto the platform. A packaged agentic solution for loan origination was announced as the first of a planned family of industry‑specific offerings.
CEO Woodson Martin underscored the company’s neutral stance. While competitors like SAP, ServiceNow, and Salesforce embed their orchestration tools within proprietary ecosystems, Martin said OutSystems “doesn’t own the data.” He highlighted that the platform has historically acted as a glue between commercial off‑the‑shelf solutions, enabling enterprises to remain independent rather than becoming an SAP or Salesforce enterprise.
A recurring theme at the conference was the rise of shadow AI. Azevedo described how internal teams can build a handful of large agentic workflows that solve company‑wide problems, yet demand from other departments is growing almost exponentially. The result is a “shadow AI” layer operating outside the central control plane, raising questions about data access and governance.
Token costs surfaced as a practical concern. Azevedo cited an Australian business value consultant burning $7,500 in tokens per week on Anthropic’s model—a figure that exceeded the company’s budget. Martin echoed the concern, noting that the latest reasoning‑heavy models released in early 2026 were driving up token bills. The company’s response is to give customers model flexibility: they can bring their own models, swap them without changing agent logic, and route requests through Amazon Bedrock to the most cost‑effective option.
OutSystems claims that its Enterprise Context Graph reduces token usage by grounding reasoning in a structured graph rather than raw code. The company expects enterprises to develop discipline around token spending, similar to other material expenses.
By remaining agnostic to the underlying systems of record, OutSystems positions itself as a bridge between the many emerging agent orchestration platforms, helping customers avoid managing multiple layers.
The latest product suite is available immediately on the ODC, with legacy support in early access. Future releases include the full Agentic Enterprise Orchestration service and the next‑generation Agent Workbench, which is already generally available.
In short, OutSystems has introduced an open agentic systems platform that integrates MCP and A2A protocols, offers a cloud‑native and legacy‑supporting developer experience, and expands its orchestration capabilities with guardrails and semantic search. Its neutral stance and focus on token management aim to address the growing demand for enterprise AI while keeping costs under control.
The centerpiece is the OutSystems Agent Experience, a platform layer that exposes two open protocols: the Model Context Protocol (MCP) and the Agent‑to‑Agent (A2A) services. MCP, introduced by Anthropic in 2024, lets large language models (LLMs) pull in external data and tools, while A2A—led by Google—enables distinct AI agents to talk to one another. By weaving these standards together, OutSystems claims developers can build, publish, and extend applications with third‑party coding tools such as Claude Code, Codex, Cursor, Kiro, and AWS’s spec‑centric IDE.
The first MCP‑enabled service is already live on the OutSystems Developer Cloud (ODC), OutSystems’ cloud‑native platform. Support for the older, self‑managed OutSystems 11 version was also released in early access. CIO Tiago Azevedo explained that the move gives legacy customers “the ability to use Claude or Codex or whatever to evolve my applications in O11.”
Beyond the Agent Experience, the company introduced Agentic Enterprise Orchestration, a new layer built on its Enterprise Context Graph. The graph stores real‑time organizational data and serves as a foundation for agent reasoning. The orchestration service adds guardrails, semantic search, and Amazon Bedrock support, empowering enterprises to manage agent workflows at scale.
OutSystems also previewed a modernization service that leverages AWS Transform and Kiro to migrate legacy systems—such as COBOL and Lotus Notes—onto the platform. A packaged agentic solution for loan origination was announced as the first of a planned family of industry‑specific offerings.
CEO Woodson Martin underscored the company’s neutral stance. While competitors like SAP, ServiceNow, and Salesforce embed their orchestration tools within proprietary ecosystems, Martin said OutSystems “doesn’t own the data.” He highlighted that the platform has historically acted as a glue between commercial off‑the‑shelf solutions, enabling enterprises to remain independent rather than becoming an SAP or Salesforce enterprise.
A recurring theme at the conference was the rise of shadow AI. Azevedo described how internal teams can build a handful of large agentic workflows that solve company‑wide problems, yet demand from other departments is growing almost exponentially. The result is a “shadow AI” layer operating outside the central control plane, raising questions about data access and governance.
Token costs surfaced as a practical concern. Azevedo cited an Australian business value consultant burning $7,500 in tokens per week on Anthropic’s model—a figure that exceeded the company’s budget. Martin echoed the concern, noting that the latest reasoning‑heavy models released in early 2026 were driving up token bills. The company’s response is to give customers model flexibility: they can bring their own models, swap them without changing agent logic, and route requests through Amazon Bedrock to the most cost‑effective option.
OutSystems claims that its Enterprise Context Graph reduces token usage by grounding reasoning in a structured graph rather than raw code. The company expects enterprises to develop discipline around token spending, similar to other material expenses.
By remaining agnostic to the underlying systems of record, OutSystems positions itself as a bridge between the many emerging agent orchestration platforms, helping customers avoid managing multiple layers.
The latest product suite is available immediately on the ODC, with legacy support in early access. Future releases include the full Agentic Enterprise Orchestration service and the next‑generation Agent Workbench, which is already generally available.
In short, OutSystems has introduced an open agentic systems platform that integrates MCP and A2A protocols, offers a cloud‑native and legacy‑supporting developer experience, and expands its orchestration capabilities with guardrails and semantic search. Its neutral stance and focus on token management aim to address the growing demand for enterprise AI while keeping costs under control.
