Stibo Systems Debuts AgentWorkx for Governed AI Agents in Master Data

Stibo Systems on September 29, 2026 announced AgentWorkx, a framework for building, deploying and governing AI agents inside master data management (MDM), along with a portfolio of agents it calls Stibo Systems Agents. The first release includes two ready-to-use agents.

Two Ways to Apply Agentic AI in STEP

The announcement gives customers two routes for applying agentic AI to MDM. They can use agents built and supported by Stibo Systems for common MDM workflows, or they can build and configure their own agents for requirements unique to their business. Both options run through the same AgentWorkx framework and within the governance, roles, permissions and enterprise controls of STEP, the Stibo Systems platform.

AgentWorkx consolidates the tools required to create and operate MDM agents into one framework: a gallery of agents, a studio for building and configuring them, orchestration, and governance. Organizations that choose to build their own agents get access to the same framework Stibo Systems’ product and engineering teams use, while those customer-built agents remain inside the platform’s controls.

First Ready-to-Use Agents and the Agents Library

The first release ships with two ready-to-use agents. Upload Anything AGT brings structured and unstructured data into the platform, and the company said it reduces onboarding processes that can take months down to days. Content Optimizer Agent generates product descriptions, SEO content and marketing copy using governed product attributes and business context. Stibo Systems said both agents are designed to turn MDM workflows into agentic experiences without requiring customers to build the underlying capabilities themselves.

The company’s agents library lists 24 available and upcoming agents organized around how data moves through STEP — getting data in, making it trustworthy and putting it to work — under what the page describes as a unified agentic governance layer. According to the library, Upload Anything guides suppliers through data ingestion, automatically maps submitted data to the target schema and taxonomy regardless of format or completeness, extracts attributes from varied file structures, validates completeness and flags gaps, cutting supplier product-data onboarding from weeks to hours.

The library describes Content Optimizer as generating channel-specific descriptions, marketing copy, SEO content and rich product detail from governed STEP attributes in any language and at catalog scale. An autonomy setting runs from human-reviewed suggestions to full auto-publish, and the library credits the agent with up to 10x faster content publishing.

Other agents listed in the library include Data Quality, which scans the catalog against completeness rules, channel specifications and consistency checks and routes lower-confidence issues to human review; Classification, which assigns products to the correct category and hierarchy automatically; Business Rules Agent, which converts natural-language requirements into working STEP business rules; GraphQL Query Writer, which generates GraphQL queries from plain-language descriptions; Entity Creation Agent, which searches MDM and other trusted sources for existing records before a new one is created; and Data Steward Copilot, which surfaces context, history and recommendations while keeping human judgment in the loop.

Platform Foundation and What Comes Next

CEO Adrian Carr tied the announcement to data reliability. “As AI agents take on more of the enterprise’s workload, the constraint is whether the data underneath can be relied on to make decisions at machine speed,” he said, adding that Stibo Systems built that foundation long before agents were a topic of discussion and that it proved to be the critical element for AI to deliver value.

Chief Product & Growth Officer Neda Nia said the AI conversation is shifting from “Can we build it?” to “Can we trust it, scale it and prove the ROI?” She said agents’ precision and performance increasingly depend on the quality of the data and context behind them, and she positioned AgentWorkx as a way for customers to innovate inside the governed environment of MDM rather than create a new generation of shadow AI.

AgentWorkx builds on MDM capabilities already in the platform, including its semantic data foundation, role-based access controls, auditability, integrations, and management of relationships across domains such as products, customers, suppliers and locations. Stibo Systems said it plans to expand both its portfolio of ready-to-use agents and the flexibility available through AgentWorkx, including work to bring customers’ preferred models into their MDM environment. Nia added that every company has data, context and institutional knowledge that is uniquely its own, and that the company’s job is to help customers put it to work without giving up trust and control.

Stibo Systems describes itself as the largest independent provider of master data management and the company behind what it calls a trusted intelligence platform. It is headquartered in Aarhus, Denmark, and operates as a subsidiary of Stibo Software Group, which the company says provides a long-term perspective through foundation ownership. Organizations interested in the framework can register early interest through the agents library.