MAS 9.2 Upgrade Planning: What Changed Since 9.1 and How to Prepare

IBM Maximo Application Suite 9.2, released in June 2026, is the second major release in the 9.x line and the most feature-rich update the platform has seen since the architectural shift from Maximo 7.6. Where MAS 9.1 laid the foundation with role-based applications, embedded AI, and the Maximo Application Framework, MAS 9.2 builds on that foundation with agentic AI workflows, a Model Context Protocol (MCP) Server for external agent integration, expanded safety and compliance capabilities, new industry solutions, and significant technical platform improvements.

For organizations currently on MAS 9.1, 9.0, or the 8.x Extended Support releases, this guide covers every major change in MAS 9.2, what it means for your environment, and how to plan your upgrade. For those still on Maximo 7.6, the direct upgrade path to 9.2 is now the most compelling reason to move, as the gap between the legacy platform and the current suite has never been wider. We will walk through the agentic AI layer, the MCP Server, the new HSE workflows, industry solution updates, technical platform changes, the rebuilt AppPoints Usage Dashboard, and a practical upgrade plan with concrete steps.

Agentic AI Workflows: From Insights to Actions

The headline feature of MAS 9.2 is the introduction of agentic AI workflows. In MAS 9.0 and 9.1, AI was primarily assistive. The Maximo Assistant answered questions, Work Order Intelligence classified work orders, and similar record detection helped users find related items. These were useful capabilities, but they stopped at providing information. A human still had to take every action.

MAS 9.2 takes a step further by introducing agentic workflows that can guide decisions and, in configured scenarios, take actions on behalf of users. The distinction between assistive AI and agentic AI is important. Assistive AI says, "Here is what I found." Agentic AI says, "Here is what I found, here is what I recommend, and here is the action I can take if you approve." The system does not operate autonomously without human oversight, but it reduces the friction between insight and execution by packaging the recommendation and the action together.

Maximo Condition Insight is the flagship agentic capability in this release. Introduced in late 2025 within the APM stack, Condition Insight brings together work orders, inspections, meter readings, and reliability strategies to identify patterns in asset behavior. In MAS 9.2, it has matured to the point where it can recommend specific corrective actions in natural language. For example, if a pump shows increasing vibration trends combined with a history of bearing failures, Condition Insight does not just flag the anomaly. It generates a recommendation that includes the specific part number to order and the timeframe for action.

The agentic layer extends to scheduling as well. Conversational scheduling, introduced in 9.2, allows planners and field service managers to explore what-if scenarios using natural language. A scheduler can ask, "What if I add two technicians to the afternoon shift next Tuesday?" or "What happens if I prioritize the boiler inspection at Plant 2 over the scheduled PMs?" The system applies the organization's constraints, technician availability, skill qualifications, travel time, and work order priority to return optimized assignment options. This replaces the manual process of adjusting assignments, running calculations, and reverting changes that has dominated scheduling workflows for years.

The Maximo Assistant has also been enhanced in 9.2. On the desktop, the assistant now supports more complex analytical queries. A user can ask for average completion times across a plant over six months, and the assistant will query the database, perform the calculation, and present the result in a formatted response with a chart. On mobile, the assistant helps technicians find asset information and review work order history using natural language, which compresses the time spent searching for context before starting a job.

The practical implication for upgrade planning is that agentic AI requires the AI Service to be deployed and configured. If your MAS 9.1 environment already has the AI Service running for the Maximo Assistant, the 9.2 upgrade will extend those capabilities automatically. If you have not yet deployed the AI Service, you will need to do so to take advantage of the agentic workflows. The AI Service runs as an additional pod in your OpenShift cluster and requires GPU resources for model inference. Plan your cluster sizing accordingly, and verify that your OpenShift cluster has sufficient GPU capacity before initiating the upgrade.

# Example: AI Service configuration in ManageWorkspace CR
apiVersion: mas.ibm.com/v1
kind: ManageWorkspace
metadata:
  name: manage-workspace
  namespace: mas-inst1-manage
spec:
  settings:
    aiService:
      enabled: true
      provider: "watsonx"
      model: "maximo-assistant-v1"
    workOrderIntelligence:
      enabled: true
      model: "maximo-failure-code-v2"
      confidenceThreshold: 0.65
      maxRecommendations: 3
      feedbackLoop: true
    conditionInsight:
      enabled: true

MCP Server: Bringing External Agents Into Maximo

One of the most architecturally significant additions in MAS 9.2 is the Model Context Protocol (MCP) Server. MCP is an open standard that allows AI agents and applications to communicate with external systems through a standardized interface. By implementing an MCP Server, IBM has made it possible for organizations to bring their own AI agents and integrate them directly with Maximo Manage APIs.

This is a bigger deal than it might sound. Since the rise of large language models and AI agents, organizations have been building custom AI tools for maintenance, operations, and reliability. These tools typically interact with Maximo through REST APIs, which works but requires custom integration code for each agent. The MCP Server standardizes this interaction. An external agent can discover available Maximo API operations, invoke them, and receive structured responses without custom integration work.

For example, an organization might have a custom AI agent that monitors weather patterns and predicts which assets are likely to be affected by approaching storms. With the MCP Server, that agent can query Maximo for assets in the predicted impact zone, check their current condition scores from Health, and automatically generate work orders for preventive inspections. All of this happens through the standardized MCP interface, without custom API integration code.

The MCP Server also opens the door for multi-agent orchestration. An organization could have one agent focused on reliability analysis, another on inventory optimization, and a third on scheduling. These agents can coordinate through the MCP Server, each contributing their specialized analysis to a unified workflow. IBM has indicated that the MCP Server is the first step in a broader strategy to make Maximo a platform that AI agents can operate on, rather than just a system that has AI features.

For administrators, the MCP Server is configured through the MAS admin interface. You control which API operations are exposed to external agents, set authentication and authorization policies, and monitor agent interactions through audit logs. The configuration is role-based, so different agents can be granted different levels of access depending on their function. A read-only analytics agent might only need query permissions, while a workflow automation agent might need work order creation and status update permissions.

From a security perspective, the MCP Server inherits the existing MAS security model. All agent interactions are authenticated through the same identity provider that powers user authentication, whether that is LDAP, SAML, OIDC, or SCIM. This means no new authentication infrastructure is required, but it also means that agent identities need to be managed with the same rigor as user identities. Create dedicated service accounts for each external agent, apply least-privilege access policies, and review agent activity logs regularly.

AppPoints Usage Dashboard and Licensing Improvements

A rebuild of the AppPoints Usage Dashboard shipped with MAS 9.2, answering a question every Maximo admin eventually gets asked: are we actually within licence, and where is usage trending? The dashboard reconstructs licence visibility from the ground up, letting admins see current usage against entitlement at a glance, drill from a multi-instance overview down to individual capability and user level, and get early, contextual alerts on where to investigate before a breach becomes a problem.

The dashboard unifies the experience across SaaS and on-prem deployments. It surfaces peak usage trends with exact timestamps rather than vague averages, so teams can act on causes instead of just reporting numbers. For organizations managing multiple MAS instances, the multi-instance overview provides a single pane of glass for license consumption across the entire estate.

MAS 9.2 also introduces a quieter Global Default Application setting under User Profile preferences, the direct result of a customer-submitted IBM Idea. Users can now specify which application opens by default when they log in, reducing friction for users who spend most of their time in a single application. This is a small change with a meaningful daily impact, especially for technicians and inspectors who previously had to navigate to their preferred application after every login.

Additional admin improvements include User Management and Security UI improvements, improved Data Import capabilities, enhanced electronic signature support for LDAP, OAuth2 SMTP authentication for outgoing emails, and support for unsupported languages through Graphite Bidi Support. The SAP Connector has been updated to use CPI (Cloud Platform Integration), and a new Oracle Fusion connector has been added.

Technical Platform Changes and Database Support

MAS 9.2 brings a substantial set of technical platform updates that upgrade planners need to factor into their preparation. The platform now runs on Java 25 (with Java 21 retained for MREF), which means your OpenShift nodes need a compatible Java runtime. BIRT has been upgraded to version 4.21, and Cognos has been updated to 12.1. MongoDB 8.0 is now supported, and CP4D 5.2 is supported for organizations running the full Cloud Pak for Data stack.

Database support has been expanded significantly. DB2 12.1 is now supported, and Oracle 26 AI support has been added once on-prem availability is confirmed. These database upgrades are not mandatory for the 9.2 upgrade itself, but they allow organizations to modernize their database infrastructure as part of the upgrade project. If you are currently running DB2 11.5 or Oracle 19c, you can upgrade to 9.2 on your current database and plan the database upgrade as a separate phase.

The Feature Channel delivery model, introduced in MAS 9.0, continues in 9.2. Feature Channels allow IBM to deliver incremental updates between major releases without requiring a full upgrade. The March, April, and June 2026 feature channels delivered updates to MAS Core 9.2.x, while the 9.0.x track received updates through May 2026. This means organizations on 9.0 can receive incremental features without immediately upgrading to 9.2, though the 9.0 track will eventually reach end of support.

For sizing and infrastructure planning, the key changes are the AI Service pod requirements and the updated MongoDB version. If you are running MongoDB 6.0, you will need to plan a MongoDB upgrade as part of the 9.2 migration. IBM provides migration tooling, but the MongoDB upgrade adds time to the overall project. Test the MongoDB migration in a non-production environment first, and verify that all applications in your MAS suite are compatible with MongoDB 8.0 before proceeding.

Key technical dependencies for MAS 9.2:
- Java 25 (Java 21 for MREF)
- BIRT 4.21
- Cognos 12.1
- MongoDB 8.0
- CP4D 5.2
- DB2 12.1 or Oracle 26 AI
- OpenShift 4.14+ recommended
- GPU resources required for AI Service

New Industry Solutions and HSE Workflows

MAS 9.2 expands the industry solution catalog with new capabilities for asset-centric waste management, contractor safety oversight, and mobile-first safety workflows. Maximo for HSE now integrates with the Envizi Greenhouse Gas Calculation API, so a fugitive emission logged in the Incidents application can be converted straight into a GHG figure using a built-in library of over 600 chemical elements. This integration is included at no additional cost for organizations running both Maximo for HSE and Envizi.

The Software Tracking application gains improved cloud-based compliance oversight for software lifecycle governance. A new Service Offerings application lets any business unit design and publish service catalog entries, not just administrators. Service bundling in Self Serve lets users submit multiple related requests in one go instead of repeating the process. The old Self Service Center application is officially deprecated in favor of Self Serve, so organizations still using it should plan their migration.

Asset Investment Planning 9.2 connects optimized investment plans directly to work execution through a continuous planning process. Key additions include publishing optimized scenarios to a centralized Investment Plan for execution and ongoing refinement, faster optimization for large concurrent asset populations, automatic intervention cost calculation through job plan integration, richer financial modelling with inflation and discount rates, and deeper scenario comparison with KPI insights and lifecycle cost analysis.

For the Spatial application, 9.2 adds 3D visualization, smart geofencing, linear visual control, heatmaps, external data dynamic upload, and data synchronization diagnostics. These spatial capabilities are particularly relevant for utilities and transportation organizations that need to visualize linear assets and manage geographically distributed infrastructure.

Practical Upgrade Plan: From 9.1 or 8.x to 9.2

Upgrading to MAS 9.2 is a project that requires careful planning, regardless of your starting point. The upgrade from 9.1 is the most straightforward, as the architectural foundation is the same. The upgrade from 8.x is more involved, as it includes the transition from the Work Center UI model to the MAF-based role-based application model. The upgrade from 7.6 is a full migration, not an upgrade, and requires a separate project plan.

Here is a recommended phased approach for organizations on 9.1:

Phase 1: Assessment (4-6 weeks). Inventory your current MAS environment, including all applications, customizations, integrations, and infrastructure. Identify which AI capabilities you want to enable and verify GPU capacity. Review the 9.2 release notes in detail and identify features that require configuration changes. Audit your Work Center customizations if any remain from 8.x, as they will need to be rebuilt on MAF.

Phase 2: Infrastructure Preparation (2-4 weeks). Upgrade OpenShift to the minimum supported version. Upgrade MongoDB if you are below 8.0. Provision GPU resources for the AI Service if not already in place. Verify database compatibility and apply any required database patches.

Phase 3: Non-Production Upgrade (3-4 weeks). Upgrade your development and test environments. Validate all integrations, customizations, and workflows. Enable the AI Service and test the agentic AI features with representative data volumes. Test the MCP Server configuration if you plan to use external agents. Conduct user acceptance testing with the role-based applications.

Phase 4: Production Upgrade (1-2 weeks plus cutover). Schedule a maintenance window. Run the upgrade with a rollback plan. Validate critical workflows immediately after upgrade. Enable new features incrementally rather than all at once, starting with the highest-value capabilities for your organization.

For organizations on 8.x, add an additional 4-6 weeks for Work Center to MAF migration and for testing the role-based application model with your user community. For 7.6 organizations, treat this as a full implementation project with a 6-12 month timeline, not an upgrade.

Practical Implications

The upgrade to MAS 9.2 is not just a technical exercise. The agentic AI capabilities, MCP Server, and expanded industry solutions change how users interact with the platform and what business processes can be supported. Organizations that upgrade without enabling the AI features will get the platform improvements but miss the strategic value. Plan to enable at least the Maximo Assistant and Condition Insight within the first 90 days of the upgrade, even if initially limited to a pilot group of users. The AppPoints Usage Dashboard should be enabled immediately, as it provides visibility that every admin needs. The MCP Server is optional for most organizations in the short term but should be evaluated if you have custom AI initiatives or plan to integrate Maximo with enterprise AI platforms.

Bottom Line

MAS 9.2 is the most significant Maximo release in years, and it deserves more than a routine upgrade project. The agentic AI layer, MCP Server, and AppPoints Dashboard collectively represent a shift in what the platform can do and how it integrates with the broader enterprise AI ecosystem. Organizations on 9.1 should plan a 3-4 month upgrade project. Organizations on 8.x or 7.6 should use 9.2 as the catalyst for their migration, as the gap between the legacy platform and the current suite is now substantial enough that delaying further means missing measurable operational value. Enable the AI features, deploy the MCP Server, and start building the workflows that will define the next phase of your asset management program.