AI Built for Asset Management: Inside Maximo 9.2's AI Revolution

In June 2026, IBM announced Maximo Application Suite 9.2, a release that fundamentally changes how AI is applied to asset management. Rather than treating AI as a separate layer or an add-on module, Maximo 9.2 embeds AI directly into the workflows that reliability, maintenance, field service, safety, and operations teams use every day. This is not AI for AI's sake. It is AI applied where work actually happens.

The release introduces several major AI capabilities: Maximo Condition Insight for agentic AI-driven asset performance management, the Maximo Assistant for natural language interaction on mobile devices, AI-enabled conversational scheduling for field service, and the MCP Server for integrating custom AI agents with Maximo APIs. Together, these capabilities represent the most significant AI investment in Maximo's history.

This article provides a deep dive into each of these capabilities, explains how they work under the hood, and offers practical guidance for organizations planning to adopt AI in their Maximo deployments.

Maximo Condition Insight: Agentic AI for Asset Performance

The centerpiece of the MAS 9.2 AI release is Maximo Condition Insight, a new AI-powered capability within the Maximo Asset Performance Management (APM) portfolio. Condition Insight uses agentic AI to interpret asset data, explain asset condition, highlight emerging trends, and recommend corrective actions. It is designed to help reliability teams move from reactive to prescriptive maintenance without requiring data science expertise.

How Condition Insight Works

Condition Insight is powered by IBM watsonx and operates across multiple data sources within the Maximo ecosystem. It evaluates work orders, inspection results, meter readings, time-series sensor data, Failure Mode and Effects Analysis (FMEA) records, and alerts to build a comprehensive picture of each asset's condition.

The process works in four steps:

1. Data aggregation: Condition Insight pulls data from across the Maximo ecosystem, including Manage work orders, Health condition indicators, Predict model outputs, and external sensor data streams.
2. Pattern analysis: The AI analyzes the aggregated data to identify patterns that indicate emerging issues. For example, it might detect that a pump's vibration levels have been increasing over the past three maintenance cycles, even though each individual reading was within normal limits.
3. Condition assessment: Based on the pattern analysis, Condition Insight produces a clear, explainable summary of the asset's condition. The output is in plain language, not technical model outputs. A reliability engineer sees: "Pump P-101 shows early signs of bearing degradation. Vibration levels have increased 12% over the last 90 days. Recommended action: schedule bearing inspection within the next 30 days."
4. Action recommendation: Condition Insight recommends specific corrective actions based on the FMEA data and historical maintenance effectiveness. The recommendations are actionable and prioritized by risk level.

The Agentic AI Difference

What makes Condition Insight different from traditional predictive maintenance is its agentic approach. Rather than simply flagging anomalies and letting humans figure out what to do, Condition Insight acts as an AI agent that reasons about asset condition, considers multiple data sources, and recommends a course of action.

For example, consider a centrifugal pump that is showing elevated vibration readings. A traditional predictive maintenance system might flag the vibration anomaly and leave it to the reliability engineer to investigate. Condition Insight goes further: it correlates the vibration data with the pump's operating history, checks the FMEA to identify likely failure modes, reviews recent work orders to see if any related maintenance has been performed, and then produces a recommendation that accounts for all of these factors.

The agentic approach also means that Condition Insight can learn from outcomes. If a reliability engineer accepts a recommendation and the resulting maintenance is successful, Condition Insight reinforces the reasoning that led to that recommendation. If a recommendation is rejected or leads to a poor outcome, the model adjusts its reasoning accordingly.

Practical Applications

Condition Insight is designed for several common use cases:

• Fleet-wide asset health monitoring: Condition Insight can monitor hundreds or thousands of assets simultaneously, flagging only those that require attention. This reduces the cognitive load on reliability teams and ensures that no asset is overlooked.
• Root cause analysis support: When an asset fails, Condition Insight can analyze the historical data leading up to the failure and identify contributing factors. This accelerates root cause analysis and helps prevent similar failures.
• Maintenance strategy optimization: By analyzing the effectiveness of past maintenance actions, Condition Insight can recommend adjustments to preventive maintenance strategies. For example, it might recommend extending the interval for a particular PM task if historical data shows it has never prevented a failure.

Data Requirements and Preparation

Condition Insight requires a solid data foundation to deliver accurate results. Organizations should ensure they have the following data elements in place before deploying:

• Complete work order history: At least 12-24 months of work order data with accurate failure codes, cause codes, and resolution descriptions
• FMEA records: Failure mode and effects analysis for critical assets, including failure modes, causes, effects, and recommended actions
• Meter and sensor data: Time-series data from meters, sensors, or IoT devices, with consistent units and timestamps
• Asset hierarchy: A well-defined asset hierarchy that reflects the physical and functional relationships between assets
• Maintenance plans: Current preventive maintenance plans with task descriptions, frequencies, and assigned crafts

Organizations that lack any of these data elements should prioritize data collection and cleansing before deploying Condition Insight. The quality of the AI outputs is directly proportional to the quality of the input data.

Maximo Assistant: Natural Language for Field and Office

The Maximo Assistant brings natural language interaction to both the desktop and mobile Maximo experience. Powered by watsonx, the assistant allows users to find asset information, review history, and complete work using plain language rather than navigating through menus and screens.

On Mobile: AI in the Field

The Maximo Assistant on Mobile is one of the most practical AI features in MAS 9.2. Field technicians can use natural language to interact with Maximo while keeping their hands free for the work at hand. Common use cases include:

• "Show me the maintenance history for Pump P-101"
• "What are the safety procedures for this work order?"
• "Record that I replaced the bearing on WO-2026-001"
• "Find the nearest inventory location for a 3-inch gate valve"

The assistant uses speech-to-text and natural language processing to understand the technician's request, queries the appropriate Maximo APIs, and returns the information in a spoken or displayed format. The entire interaction happens within the Maximo Mobile app, with no need to switch between applications.

On Desktop: Conversational Analytics

On the desktop, the Maximo Assistant supports more complex queries and analytics. Reliability engineers can ask questions like:

• "Which assets have the highest maintenance cost per unit of production?"
• "Show me work orders that are past due by more than 30 days"
• "What is the mean time between failures for our centrifugal pumps?"
• "Compare the reliability of Vendor A and Vendor B equipment"

The assistant translates these natural language queries into Maximo database queries, executes them, and returns the results in a readable format. This makes Maximo's data accessible to users who are not familiar with the database schema or reporting tools.

Implementation Considerations

Deploying the Maximo Assistant requires some preparation. The assistant uses watsonx's natural language processing capabilities, which need to be configured with your organization's specific terminology and asset naming conventions. For example, if your organization uses internal codes for equipment types, the assistant needs to learn those mappings to understand queries like "Show me all PMs for HVA-101."

IBM provides a configuration toolkit that allows organizations to train the assistant on their specific vocabulary. The training process involves uploading a glossary of terms, sample queries, and expected responses. The more training data you provide, the more accurate the assistant will be.

AI-Enabled Field Service Management

MAS 9.2 brings AI-powered intelligence to Field Service Management (FSM) in several important ways. The most significant is AI-enabled conversational scheduling, which allows planners and dispatchers to interact with the scheduling engine using natural language.

Conversational What-If Analysis

Planners can ask the scheduling engine questions like:

• "What happens if I add three more emergency work orders to today's schedule?"
• "Can I prioritize all work at the Smithfield plant?"
• "Show me the impact of moving the PM-102 inspection to next week"

The AI engine evaluates the request against current constraints (technician availability, skill requirements, travel time, parts availability, regulatory deadlines) and returns a proposed schedule. The planner can accept, modify, or reject the proposal and ask follow-up questions.

Intelligent Technician Assignment

The AI scheduling engine also optimizes technician assignment based on skills, certifications, location, and workload. When a new work order is created, the system automatically identifies the best-qualified technician who is available and nearby, and proposes the assignment. Dispatchers can override the recommendation if needed, but the AI handles the routine assignments, freeing dispatchers to focus on exceptions and complex scenarios.

The optimization algorithm considers multiple factors simultaneously:

• Skill match: Does the technician have the required certifications and experience?
• Location: How far is the technician from the work site?
• Availability: Is the technician free during the required time window?
• Workload: Does the technician have capacity to take on additional work?
• Priority: Is this work order urgent or routine?

By optimizing across all of these factors, the AI can produce schedules that are more efficient than manual scheduling, particularly for large field service operations with dozens or hundreds of technicians.

Visual Inspection with On-Device AI

Maximo Visual Inspection (MVI) in MAS 9.2 supports AI-based visual inspection with local inference directly on the device. This means that field technicians can use their mobile device's camera to inspect assets, and the AI model runs locally on the device without requiring a network connection. The model can identify defects, corrosion, cracks, and other visual anomalies, and automatically create inspection records in Maximo.

The on-device inference is a critical feature for field environments with limited connectivity. Technicians working in remote locations, underground facilities, or areas with poor cellular coverage can still use AI-powered visual inspection. The inspection results are synced to Maximo when connectivity is restored.

Training the visual inspection models is done through the MVI training interface, where users upload labeled images of defects and normal conditions. IBM provides pre-trained models for common inspection scenarios, including corrosion detection, crack detection, and equipment identification. Organizations can also train custom models for their specific equipment types.

MCP Server: Bring Your Own AI Agents

One of the most forward-looking features in MAS 9.2 is the MCP Server, which exposes Maximo Manage APIs through the Model Context Protocol (MCP). This allows organizations to bring their own AI agents and integrate them directly with Maximo, enabling AI to participate in operational processes without relying on manual coordination between disconnected tools.

What the MCP Server Enables

The MCP Server provides a standardized interface for AI agents to interact with Maximo business objects. An AI agent can:

• Query work orders, assets, locations, and inventory
• Create and update work orders
• Trigger workflows and approval processes
• Retrieve maintenance history and inspection records
• Access FMEA data and reliability strategies

This opens up a wide range of possibilities for custom AI applications. For example:

• A safety AI agent that monitors work order descriptions for safety risks and automatically adds required safety plans
• A compliance AI agent that reviews work order completion data against regulatory requirements and flags gaps
• A supply chain AI agent that monitors inventory levels and automatically creates purchase requisitions when stock falls below reorder points
• A reliability AI agent that analyzes work order data and recommends updates to preventive maintenance strategies

Integration Architecture

The MCP Server is deployed as a containerized service within the MAS cluster. It authenticates using the same identity and access management framework as other MAS components, ensuring that AI agents operate within the same security boundaries as human users. Each AI agent is assigned a service account with specific permissions, so organizations can control exactly what each agent can access and modify.

The protocol is based on standard REST APIs with JSON payloads, making it accessible to any AI framework that supports HTTP-based tool calling. IBM provides client libraries and examples for popular AI frameworks, including LangChain, LlamaIndex, and custom Python agents.

Practical Implications

The AI capabilities in MAS 9.2 represent a significant step forward for asset management, but they also require organizations to prepare their data and processes. Here are the practical implications for organizations planning to adopt AI in Maximo:

1. Data quality is the foundation. AI models are only as good as the data they are trained on. Condition Insight requires clean, consistent work order data, accurate FMEA records, and reliable sensor data. Organizations with poor data quality will need to invest in data cleansing before they can realize the benefits of AI.
2. Start with a focused use case. Rather than trying to deploy all AI capabilities at once, start with a single use case where AI can deliver immediate value. Condition Insight for a critical asset fleet is a good starting point. Prove the value, learn the operational requirements, and then expand.
3. Plan for organizational change. AI changes the role of reliability engineers, planners, and technicians. Engineers shift from data gathering to decision-making. Planners shift from manual scheduling to exception management. Technicians gain new tools for field work. These changes require training, new workflows, and leadership support.
4. Consider the MCP Server for custom needs. The MCP Server is a powerful capability for organizations with unique AI requirements. If you have specific compliance, safety, or operational needs that are not addressed by the built-in AI features, the MCP Server allows you to build custom AI agents that integrate directly with Maximo.
5. Monitor and validate AI outputs. AI recommendations should be treated as decision support, not decision replacement. Organizations should establish processes for monitoring AI outputs, validating recommendations, and providing feedback to improve model performance over time.

The Bottom Line

Maximo 9.2 represents a fundamental shift in how AI is applied to asset management. By embedding AI directly into operational workflows rather than treating it as a separate capability, IBM has made AI practical and accessible for maintenance and reliability teams. Condition Insight brings agentic AI to asset performance management. The Maximo Assistant makes natural language interaction a reality for field and office workers. AI-enabled scheduling transforms field service management. And the MCP Server opens the door for custom AI agents.

The organizations that will benefit most from these capabilities are those that invest in data quality, start with focused use cases, plan for organizational change, and treat AI as a tool for augmenting human expertise rather than replacing it. The AI revolution in asset management is here. The question is not whether to adopt it, but how quickly you can prepare your organization to take advantage of it.