AI in Maximo: From Chatbots to Agentic Workflows in MAS 9.2

Artificial intelligence in Maximo has evolved rapidly over the past two years. What began as a simple chatbot for answering maintenance questions has transformed into a sophisticated ecosystem of AI capabilities embedded directly into asset management workflows. The release of Maximo Application Suite 9.2 in July 2026 marks a significant milestone in this evolution, introducing agentic AI capabilities that can reason about asset condition, recommend actions, and even execute tasks across systems.

This article provides a comprehensive technical overview of AI in Maximo as of mid-2026. We will cover the Maximo AI Assistant and its maturation beyond a simple chatbot, the new Condition Insight capability for agentic asset performance management, the MCP Server that enables external AI agents to participate in Maximo workflows, the practical realities of predictive maintenance and visual inspection, and a framework for building an AI roadmap for your Maximo deployment. Whether you are just starting to explore AI or looking to expand existing capabilities, this guide will help you understand what is possible and how to get started.

The Maximo AI Assistant: Beyond the Chatbot

The Maximo AI Assistant has been available since MAS 8.x, but its capabilities have expanded dramatically in the 9.x releases. In MAS 9.2, the AI Assistant has matured well beyond a simple question-answering chatbot into a multi-functional assistant that can help with configuration guidance, data retrieval, reporting, and asset condition insights.

The AI Assistant is powered by IBM watsonx and integrates directly with the Maximo data model. This means it understands the relationships between assets, work orders, inventory, purchase orders, and other Maximo objects. When a maintenance technician asks "What is the status of work order WO-2026-00421?", the AI Assistant does not just search for a text match. It queries the Maximo data model, understands the context of the question, and returns a precise answer with relevant details such as the current status, assigned craft, scheduled dates, and any associated safety requirements.

The AI Assistant supports natural language queries across several domains:

• Work order intelligence: "Show me all open work orders for the Bedford site with priority 1 that are overdue."
• Asset condition insights: "What is the current condition of cooling tower CT-003 and when was it last inspected?"
• Reporting: "Generate a monthly report of corrective maintenance costs by asset type for Q2 2026."
• Configuration guidance: "How do I set up a new Publish Channel for work order status changes?"
• Inventory queries: "What is the current stock level of 4-inch gate valves at the Houston warehouse?"

The AI Assistant is available through the Maximo web interface and, critically for field workers, through Maximo Mobile. The mobile integration means technicians can ask questions using natural language while standing next to the asset they are working on, without needing to navigate through multiple screens on a small mobile device. This is a significant usability improvement that reduces the time technicians spend searching for information and increases the time they spend actually working on assets.

A key architectural detail is that the AI Assistant operates within the Maximo security model. It only surfaces data that the requesting user has permission to see. This means a technician will not see financial data or sensitive information that is outside their role, even if they ask for it. The AI Assistant respects all existing Maximo security configurations, including site-level security, application-level security, and data-level restrictions. This security integration is often overlooked but is critical for enterprise deployments where data access must be tightly controlled.

Condition Insight: Agentic AI for Asset Performance Management

The most significant AI announcement in the Maximo ecosystem for 2025-2026 is Maximo Condition Insight, a new AI-powered capability within the Maximo Application Suite that enables asset-intensive organizations to gain instant, explainable insights to support condition-based maintenance. Condition Insight is an agentic AI capability within Maximo APM that works in concert with other applications and AI capabilities that are part of the broader Maximo Application Suite.

What makes Condition Insight different from traditional analytics tools is its agentic nature. Rather than simply displaying dashboards or generating reports, Condition Insight actively interprets asset data, explains asset condition, highlights emerging trends, and recommends corrective actions. It is designed to answer the question that every maintenance team asks: "What should I do about this asset?" This shift from passive reporting to active recommendation is what makes Condition Insight a genuinely new capability rather than an incremental improvement.

Condition Insight evaluates multiple data sources to assess asset condition:

• Work orders: Historical and open work orders provide context about recurring issues, repair frequency, and maintenance history.
• Meter readings: Time-series data from meters tracks usage, runtime, and performance degradation.
• Time-series data: IoT sensor data provides real-time condition monitoring for vibration, temperature, pressure, and other parameters.
• Failure Mode and Effects Analysis (FMEA): Pre-defined failure modes and their effects help Condition Insight understand what could go wrong and what the consequences would be.
• Alerts: System-generated alerts from Monitor and other monitoring tools provide early warning of emerging issues.

Condition Insight processes these data sources using watsonx AI models and returns a clear, explainable summary of asset condition, trends, and recommended actions, all communicated in plain, understandable language. The output is designed to be actionable by maintenance teams without requiring data science expertise.

A practical example of how Condition Insight works:

Asset: Cooling Tower CT-003
Condition: Degraded (Confidence: 87%)
Trend: Vibration levels increasing 12% month-over-month for 3 months
Likely Cause: Bearing wear on fan motor assembly
Recommended Action: Schedule bearing replacement within 14 days
Supporting Evidence:
  - Vibration reading 7.8 mm/s exceeds threshold of 6.0 mm/s
  - Similar failure mode (FMEA-042) resulted in unplanned outage at 9.2 mm/s
  - Last bearing replacement: 18 months ago (expected life: 24 months)
  - Two corrective work orders for vibration in past 6 months

The explainability aspect is critical. Maintenance teams need to understand why the AI is making a recommendation before they will act on it. Condition Insight provides the reasoning behind every recommendation, including the specific data points and failure modes that drove the analysis. This transparency builds trust and enables human-in-the-loop decision-making. Without explainability, AI recommendations are just black-box suggestions that teams will ignore.

Condition Insight was released as part of the AI Service Component feature release 9.2.0 in December 2025 and is now generally available with MAS 9.2. It represents IBM's strategic direction for asset performance management: moving from descriptive analytics (what happened) to diagnostic analytics (why it happened) to prescriptive analytics (what to do about it).

The MCP Server: Bringing External AI Agents into Maximo

One of the most architecturally significant features in MAS 9.2 is the MCP (Model Context Protocol) Server. The MCP Server enables organizations to bring their own AI agents and integrate them directly with Maximo Manage APIs, allowing AI to participate in operational processes without relying on manual coordination between disconnected tools.

The MCP Server acts as a translation layer between AI agents and Maximo. Traditional REST APIs were built for developers who understand HTTP methods and JSON payloads. AI agents, on the other hand, reason in terms of business objectives. The MCP Server translates "create a work order for the cooling tower vibration issue" into the correct system interactions, without the AI needing to know Maximo's endpoint structure, authentication requirements, or data schemas.

This is a fundamentally different approach to integration. Instead of building custom code to connect AI agents to Maximo, organizations can use the MCP Server to expose Maximo capabilities to any AI agent that supports the MCP protocol. This includes agents built on watsonx, but also agents built on other AI platforms that support the open MCP standard. The open nature of the protocol means organizations are not locked into a single AI vendor.

The MCP Server supports several categories of agent interactions:

• Data retrieval: Agents can query Maximo for asset information, work order status, inventory levels, and other data.
• Action execution: Agents can create work orders, update statuses, assign resources, and trigger workflows.
• Orchestration: Agents can coordinate multi-step processes that span Maximo and other systems.
• Notification: Agents can subscribe to events and receive notifications when specific conditions are met.

Crucially, the MCP Server does not replace REST APIs or enterprise integration platforms. It is an additional integration pathway designed specifically for AI agent interactions. A single MCP Server cannot carry a complex, multi-system business process on its own. Orchestration layers are still essential for processes that span multiple enterprise systems. But for AI-to-Maximo interactions, the MCP Server dramatically reduces the integration effort.

The security model for the MCP Server is built on the same foundation as the MAS REST API. All agent interactions are authenticated using OAuth 2.0, and agents are subject to the same role-based access controls as human users. This means an AI agent cannot access data or perform actions that the agent's configured identity does not have permission to access. Audit logging captures all agent interactions, providing a complete record of what the AI did and when.

Predictive Maintenance and Visual Inspection: The Proven AI Workhorses

While Condition Insight and the MCP Server represent the cutting edge of AI in Maximo, the proven workhorses of AI-powered maintenance remain predictive maintenance (via Maximo Predict) and visual inspection (via Maximo Visual Inspection). These capabilities have been available for several years and have the deepest track record of real-world results.

Maximo Predict uses machine learning models to analyze historical work order data, meter readings, and sensor data to predict when an asset is likely to fail. The models are trained on the organization's own data, learning the specific failure patterns that are relevant to their assets and operating conditions. Predict generates a remaining useful life estimate for each asset, along with a confidence score and the key factors driving the prediction.

The practical application of Predict is straightforward: assets with a low remaining useful life and high criticality are prioritized for proactive maintenance. This shifts the maintenance strategy from time-based (replace every 12 months) to condition-based (replace when the model predicts failure is imminent). The result is fewer unnecessary replacements, fewer unexpected failures, and optimized maintenance resource allocation. Organizations typically see a 20-40% reduction in unplanned downtime after deploying Predict on their critical asset classes.

Maximo Visual Inspection uses deep learning to analyze images and videos for classification and defect detection. The solution helps users train, validate, and deploy AI models without requiring data science expertise. The workflow is designed for domain experts, not data scientists: a maintenance supervisor can upload a set of labeled images (good condition vs. cracked vs. corroded), train a model with a few clicks, and deploy it to inspection workflows.

The integration between Visual Inspection and the broader Maximo platform is what makes it powerful. When an inspection identifies a defect, the system can automatically create a work order, attach the inspection images as evidence, and route the work order to the appropriate maintenance team. This closed-loop integration ensures that inspection findings are acted on, not just documented. In MAS 9.2, Visual Inspection also supports local inference directly on mobile devices, enabling AI-powered inspections even in locations without network connectivity.

A critical insight from practitioners is that the quality of the training data is the single most important factor in the success of both Predict and Visual Inspection. As Matt Boehne explains in his analysis of AI in Maximo, "AI is rooted in two things: data and algorithms. The algorithms have to work with the type of data that you are feeding into it. And you have to have good data in order to get good output from the algorithms." Organizations that invest in data quality before deploying AI see significantly better results than those that rush to deploy AI on poor-quality data. This is not a step that can be skipped.

Building an AI Roadmap for Your Maximo Deployment

Given the range of AI capabilities now available in Maximo, the challenge for most organizations is not finding AI use cases, but prioritizing them and building a coherent implementation roadmap. Based on the experiences of early adopters, a phased approach is recommended.

Phase 1: Foundation (3-6 months). Start with the AI Assistant to give your team experience with natural language interfaces. Deploy Maximo Visual Inspection for a single, high-value inspection use case. Focus on data quality: clean up asset data, standardize work order descriptions, and ensure meter readings are accurate and complete. The goal of this phase is to build organizational confidence in AI and establish the data foundation for more advanced capabilities.

Phase 2: Expansion (6-12 months). Deploy Maximo Predict for your most critical asset classes. Start with assets where failure has the highest consequences: critical production equipment, safety-critical systems, or assets with long lead times for replacement parts. Expand Visual Inspection to additional use cases. Begin experimenting with Condition Insight in a non-production environment. The goal of this phase is to demonstrate measurable ROI from AI-powered maintenance.

Phase 3: Advanced (12-24 months). Deploy Condition Insight for production use across your asset portfolio. Implement the MCP Server and begin integrating external AI agents into Maximo workflows. Explore agentic workflows that combine multiple AI capabilities: for example, an agent that monitors Condition Insight recommendations, checks inventory for required parts, schedules the work, and creates the work order automatically. The goal of this phase is to achieve autonomous maintenance operations for appropriate use cases.

Practical Implications

The AI capabilities in MAS 9.2 represent a genuine leap forward, but they also require organizations to invest in the foundations that make AI work. Data quality is the most critical success factor. AI models are only as good as the data they are trained on, and poor data quality will produce unreliable recommendations that erode trust in the system. Before deploying any AI capability, conduct a data quality assessment and address the gaps you find.

Organizations should also plan for organizational change. AI-powered maintenance changes the role of maintenance technicians from reactive problem-solvers to proactive asset managers. This requires training, new skills, and a shift in mindset. The organizations that invest in change management alongside the technology deployment see faster adoption and better results. Do not underestimate the human side of the AI transformation.

Security and governance are also important considerations. The MCP Server and agentic AI capabilities introduce new attack surfaces and new governance requirements. Organizations should extend their existing Maximo security model to cover AI agent interactions, including authentication, authorization, audit logging, and model governance. Establish clear policies for what AI agents are allowed to do autonomously versus what requires human approval.

The Bottom Line

AI in Maximo has evolved from a single chatbot to a comprehensive ecosystem spanning natural language interaction, agentic asset performance management, external AI agent integration, predictive maintenance, and visual inspection. MAS 9.2 represents the most significant AI release in Maximo's history, with Condition Insight and the MCP Server opening up capabilities that were not available in any previous version.

The strategic message for Maximo practitioners is that AI is no longer a future capability to plan for. It is here, it is production-ready, and it is delivering measurable results for organizations that invest in the data foundations and organizational change required to make it work. The question is no longer whether to adopt AI in Maximo, but where to start and how fast to move. The organizations that start now, even with a small pilot, will be the ones with the experience and data foundation to scale AI across their asset portfolio when the next wave of capabilities arrives.