STIBO Systems released STEP 2026 on March 25, 2026, and it says a lot about where the platform is heading. The headline is not a new screen. It is governed master data that AI agents can reach without a custom integration project. For companies that run product, supplier, or customer data on STIBO STEP, this update changes what is practical to deliver this year.
Master data rarely gets attention in the boardroom until something breaks: a product launch slips because attributes are missing, a retailer rejects a feed, or an AI pilot gives confident answers based on stale records. The 2026 release speaks directly to those moments. Here is what changed, why it matters, and what to plan for before you update.
Where STEP 2026 Fits
STIBO now ships STEP on a quarterly cadence. The STIBO Systems product updates center lists 2026 as the latest release, following 2026.1 in March and 2026.2 in June, with 2026.4 scheduled for December 9. STIBO refers to STEP as the STIBO Systems platform, and its update pages group features by data domain and by license.
The release builds on a foundation laid in June. In 2026.2, the Consumer MCP Server became generally available, giving large language models and AI agents a governed way to discover and use master data through the open Model Context Protocol. STIBO framed that move in its MCP Server launch announcement as making MDM the context layer for enterprise AI. Version 2026 is about making that capability easier to switch on and more useful once it is running.
What Is New in STIBO STEP 2026
Self-service AI access to trusted data
Getting master data in front of AI agents used to take real effort: activating the solution, configuring publishing, and hand-building services and semantic schemas. According to the official 2026 update page, Data as a Service and the Consumer MCP Server can now be enabled from the Self-Service UI. DaaS Easy-Setup configures STEP-to-DaaS publishing from the Workbench and publishes a sample dataset, and the Semantic Data Graph generates MCP-ready services and semantic schemas from the existing data model and metadata. STIBO lists these improvements as included in the base license.
CoPilot and AI prompt creation in the New UI
Business users get two AI features, both tied to the ProductGen AI license. A CoPilot chat interface, released in Limited Availability, lets people search and compare products in natural language, ask for data quality validation, and open product links straight from the answer. AI Prompt Creation, a Phase 1 feature, lets teams create, test, and manage prompts in the New UI and apply them to product data for tasks like translation and content generation, without technical setup.
Context, translation, and relationships in one place
Global catalogs vary by language and market, and the New UI previously required switching contexts by hand. The new Multi-Context Tab, in its first phase, shows and maintains attribute values across several contexts side by side. A Translation tab brings review and approval of translations into a single view. References in the Details view are now structured and editable inline, so related products, assets, and metadata can be managed without leaving the workspace.
Syndication, onboarding, and matching at scale
On the exchange side, STEP to PDX syndication setup is simpler, with clearer entry points in the New UI and support for channel-issued external product IDs. PDX now supports search and automation rules across all language layers. Automotive companies gain PIES 8.0 support under the Auto Care license. For customer, business partner, and supplier data, the unmerge process now reapplies manually edited values automatically, and Match & Merge import throughput grows near-linearly as workers are added.
Before and After the 2026 Update
The table below summarizes how STIBO itself describes the shift, using the "what changed" notes on its update page.
| Area | Before 2026 | With 2026 |
|---|---|---|
| AI agent access to master data | Consumer MCP Server available since 2026.2, but activation, publishing setup, and schema creation took considerable effort | Enabled from the Self-Service UI, with DaaS Easy-Setup and an auto-generated Semantic Data Graph |
| Finding product data | Navigating screens and tabs, running searches, or knowing the data model | CoPilot chat for natural-language search, comparison, and validation (Limited Availability) |
| AI prompts | Required technical setup and manual configuration | Created, tested, and managed in the New UI (Phase 1) |
| Market and language variants | Manual context switching to view or compare values | Multi-Context Tab and Translation tab in one view |
| Product relationships | Data pieced together across separate views | Structured references edited inline in the Details view |
| Supplier and channel exchange | PDX search and rules in the default language only; complex syndication setup | Multi-language search and rules, simpler syndication, PIES 8.0 support |
| Golden record unmerge | Manual edits picked again from unused values | Manual edits and references reapplied automatically |
What These Features Mean for the Business
AI readiness without a side project
Most AI initiatives stall on data, not models. When agents read product or customer data from a governed source, with roles, permissions, and semantics intact, answers are more accurate and easier to defend. Self-service activation and generated schemas lower the cost of trying that approach, which means an AI use case can move from idea to pilot without a separate integration build. We covered that gap in more depth in Why AI Pilots Stall Before Production.
Faster time to market across regions
Every day a product sits waiting for translated copy or market-specific attributes is a day it is not selling. Side-by-side context editing, a dedicated translation workflow, and prompt-driven content generation shorten the path from a new SKU to a complete, localized listing. Teams also spend less time hunting through screens to find what is missing.
Better data quality with less rework
Quality problems compound. A wrong value restored after an unmerge, or a reference edited out of context, can ripple into pricing, ordering, and customer service. Automatic reapplication of manual edits and inline reference management remove two common sources of quiet errors, and CoPilot gives non-experts a direct way to ask whether a record passes validation.
Smoother supplier and channel exchange
Supplier onboarding and retailer syndication are where master data meets the outside world. Multi-language search and rules in PDX, simpler syndication setup, and PIES 8.0 support reduce the manual workarounds that slow onboarding and create compliance risk with trading partners.
AgentWorkx and the Road Ahead
In September 2026, the same month as the release, STIBO announced AgentWorkx, a framework for building, deploying, and governing AI agents inside MDM. As described in the AgentWorkx announcement, it includes a gallery of agents, a studio for configuring them, orchestration, and governance, all operating within STEP roles and permissions. The first two STIBO-built agents are the Upload Anything Agent for bringing structured and unstructured data into the platform, and the Content Optimizer Agent for product descriptions and SEO copy. STIBO presents AgentWorkx as an announcement with early-interest registration, so confirm availability and licensing with your account team before you plan around it.
Plan the Upgrade Before You Need It
A few details on the update page deserve attention from IT leaders. Version 2026 is available only on SaaS, and STIBO asks on-premises customers to start migrating. The release is supported for nine months, through June 23, 2027, and customers on Continuous Updates should have production updated by December 3, 2026.
There are also technical changes worth testing. STEP now uses the Java platform default locale data, so date, time, and number formats may shift. The Qlik and Tableau analytics widgets are deprecated, and Basic Authentication is deprecated for new PDX Syndication delivery configurations. As always, integrations, Extension API extensions, and custom business rules should be regression-tested in a preview sandbox before production moves.
How DevWise Helps With STIBO STEP
DevWise works with companies at every stage of a STIBO STEP program. We design and implement STEP solutions, shape data models that reflect how the business actually sells and buys, and build the integrations and APIs that connect STEP to ERP, commerce, PIM channels, and analytics. We plan and run migrations, including moves from on-premises to SaaS, and help teams set up data governance with clear ownership, workflows, and quality rules.
After go-live, our engineers provide ongoing support, upgrade testing for each quarterly release, and enhancements as new features like the Consumer MCP Server come into scope. Clients can work with a dedicated nearshore team in overlapping time zones, so questions get answered in the same working day. You can read more about our approach on our master data management services page, or start with our overview of master data management for enterprise teams.
If you are weighing the 2026 update, an on-premises to SaaS move, or a first AI use case on top of STEP, DevWise can help you sort out what to do first. Talk with our team about where to start.
