From Predictive to Prescriptive: The Maturation of AI Inside Maximo in 2026

For years, the promise of AI in enterprise asset management was always one release away. Maximo Predict delivered machine learning models that could forecast asset failures, but the models lived in a separate application that reliability engineers had to consciously choose to use. The predictions were accurate, but they did not appear in the work order that a planner was reviewing or the inspection form that a technician was completing in the field.

MAS 9.2, released in June 2026, changes that equation. AI is no longer a separate module that organizations opt into. It is embedded in the daily workflows that reliability, maintenance, field service, safety, and operations teams already use. IBM describes this as asset-first AI, and the distinction matters. The AI is not bolted onto the EAM platform. It is woven into the fabric of how work is managed, prioritized, and executed.

This article traces the evolution of AI capabilities inside Maximo from the foundational Predict application through the Condition Insight capability that defines MAS 9.2. It covers the five AI capabilities that organizations encounter first, the data requirements for each, and the implementation path for moving from reactive maintenance to prescriptive, AI-driven reliability.

The Five AI Capabilities in MAS 9.2

As of the MAS 9.2 release, five AI capabilities are available to organizations running the platform. Each capability addresses a different stage of the maintenance maturity curve, and each has different data requirements and implementation considerations.

Maximo Predict: The Foundation

Maximo Predict is the application that started the AI journey in Maximo, and it remains the foundation for everything MAS 9.2 builds on. Predict uses AI and machine learning to analyze historical maintenance records, operational data, inspection reports, and environmental data to predict downtime, degradation, and failures.

The implementation workflow for Predict is well-documented and follows a clear sequence:

1. Create asset groups based on asset type, criticality, or operational context. Predict works best when analyzing groups of similar assets rather than individual assets in isolation. A group might be all centrifugal pumps at the north plant or all transformers rated above 500 kVA.
2. Train predictive models using the default notebooks provided with Predict. A data scientist works with the group ID and uses Jupyter notebooks to build and train model instances. The notebooks handle the common algorithms, and the data scientist focuses on feature engineering and model selection.
3. Deploy the trained model so that it generates predictions for each asset in the group. Predictions include current failure probability and estimated failure date.
4. Monitor predictions through the Predictions section in Maximo, which shows failure probability, estimated failure date, and recommended actions for each asset.
5. Use work queues to track assets with high failure probability or assets predicted to fail before the next scheduled PM work order. These work queues become the daily planning tool for reliability engineers.

# Example: Training a Predict model for centrifugal pumps
# This runs inside the Maximo Predict Jupyter notebook environment

import maximo_predict as mp

# Define the asset group
group_id = "CENTRIFUGAL_PUMPS_NORTH_PLANT"
asset_group = mp.create_asset_group(
    name=group_id,
    description="All centrifugal pumps at North Plant",
    assets=mp.query_assets(
        asset_type="CENTRIFUGAL_PUMP",
        location="NORTH_PLANT"
    )
)

# Train the model using the default notebook
model = mp.train_model(
    group_id=group_id,
    model_type="FAILURE_PREDICTION",
    training_data=mp.get_training_data(
        group_id=group_id,
        lookback_days=730  # Use 2 years of history
    ),
    features=[
        "meter_readings",
        "work_order_history",
        "failure_codes",
        "inspection_results",
        "operating_hours"
    ]
)

# Deploy the model
mp.deploy_model(model_id=model.id)

# Check predictions for the group
predictions = mp.get_predictions(group_id=group_id)
for asset_id, prediction in predictions.items():
    print(f"Asset: {asset_id}")
    print(f"  Failure Probability: {prediction.failure_probability}%")
    print(f"  Estimated Failure Date: {prediction.estimated_failure_date}")
    print(f"  Days to Failure: {prediction.days_to_failure}")
    print(f"  Recommended Action: {prediction.recommended_action}")
    print()

The data requirements for Predict are the most demanding of the five AI capabilities. The models need at least 18 to 24 months of historical data to produce reliable predictions. The data must include consistent failure coding, meter readings at regular intervals, and work order history with actual completion data. Organizations with incomplete or inconsistent historical data will need to invest in data remediation before training models.

Maximo Condition Insight: The Agentic Layer

Maximo Condition Insight is the centerpiece of MAS 9.2's AI strategy. Introduced in late 2025 and matured through the 9.2 release, it brings together work orders, inspections, meter readings, and reliability strategies to identify patterns in asset behavior and recommend what to do next.

Condition Insight is described as an agentic AI capability, which means it does not just analyze data. It interprets it, explains what is happening in natural language, and recommends specific corrective actions. The recommendations appear in context, alongside the asset record, the work order, or the inspection form that the user is already working with.

The difference between Predict and Condition Insight is the difference between a forecast and a recommendation. Predict tells you that a pump has a 78 percent probability of failure within the next 14 days. Condition Insight tells you that the pump's bearing temperature has been trending upward for three weeks, that the vibration analysis from last week's inspection showed elevated amplitude in the high-frequency band, and that the recommended action is to schedule a bearing replacement before the next operational cycle. It also tells you that similar pumps in the same asset group have failed within 20 days of showing this pattern.

Condition Insight is powered by watsonx, IBM's enterprise AI platform, which provides the natural language processing, pattern recognition, and recommendation generation capabilities. The integration with watsonx means that Condition Insight can incorporate unstructured data (inspection notes, maintenance logs, safety reports) alongside structured data (meter readings, work order history, failure codes).

Maximo Monitor: IoT Data Ingestion at Scale

Maximo Monitor (formerly Maximo Health) provides AI-enabled remote monitoring of asset condition at scale. It ingests sensor data from IoT devices, calculates health scores, and triggers alerts when condition thresholds are exceeded.

Monitor is the data pipeline that feeds Predict and Condition Insight. Without Monitor, the other AI capabilities have to rely on manually entered meter readings and inspection data, which limits their accuracy and timeliness. With Monitor, sensor data flows continuously into the platform, enabling real-time condition assessment and near-real-time predictions.

// Example: Monitor alert configuration for a centrifugal pump
{
  "asset_id": "PUMP-1023",
  "monitoring_profile": "COOLING_WATER_PUMP",
  "sensors": [
    {
      "type": "TEMPERATURE",
      "location": "BEARING_HOUSING",
      "normal_range": "35-65°C",
      "warning_threshold": 75,
      "critical_threshold": 85,
      "sampling_interval_seconds": 60
    },
    {
      "type": "VIBRATION",
      "location": "DRIVE_END_BEARING",
      "normal_range": "0-4.5 mm/s RMS",
      "warning_threshold": 5.5,
      "critical_threshold": 7.0,
      "sampling_interval_seconds": 300
    },
    {
      "type": "CURRENT",
      "location": "MOTOR",
      "normal_range": "12-18A",
      "warning_threshold": 22,
      "critical_threshold": 26,
      "sampling_interval_seconds": 60
    }
  ],
  "alert_actions": [
    {
      "condition": "ANY_SENSOR > warning_threshold",
      "action": "CREATE_CONDITION_INSIGHT_REQUEST",
      "priority": "MEDIUM"
    },
    {
      "condition": "ANY_SENSOR > critical_threshold",
      "action": "CREATE_WORK_ORDER",
      "work_type": "EM",
      "priority": "HIGH"
    }
  ]
}

The configuration above shows how Monitor, Condition Insight, and work order creation work together. When a sensor exceeds a warning threshold, Monitor automatically triggers a Condition Insight request to analyze the asset's condition and generate a recommendation. When a sensor exceeds a critical threshold, Monitor creates an emergency work order directly. This is the closed-loop system that makes MAS 9.2's AI capabilities practical for daily operations.

Maximo Visual Inspection: Computer Vision for Defect Detection

Maximo Visual Inspection uses computer vision models to detect defects from images and video, with the option to run inference locally on mobile devices. This capability does not depend on historical data, which makes it the fastest AI capability to deploy in a new Maximo environment.

The implementation process for Visual Inspection involves training a custom model on labeled images of defects. For example, a utility company might train a model to detect rust on transmission towers by providing hundreds of images of towers with and without rust, with each image labeled to indicate the presence, location, and severity of corrosion.

# Example: Training a Visual Inspection model for rust detection
# Using the Maximo Visual Inspection API

import visual_inspection as vi

# Create a new model
model = vi.create_model(
    name="transmission_tower_rust_detection",
    description="Detects rust and corrosion on transmission towers",
    model_type="OBJECT_DETECTION"
)

# Upload labeled training images
training_data = vi.upload_labeled_images(
    model_id=model.id,
    image_directory="/data/training/tower_rust/",
    labels=["no_corrosion", "minor_corrosion", "severe_corrosion"],
    split_ratio={"train": 0.8, "validation": 0.1, "test": 0.1}
)

# Train the model
training_result = vi.train_model(
    model_id=model.id,
    epochs=50,
    learning_rate=0.001,
    augmentation=True
)

print(f"Model accuracy: {training_result.accuracy}")
print(f"Validation accuracy: {training_result.validation_accuracy}")

# Deploy the model for mobile inference
vi.deploy_model(
    model_id=model.id,
    target="MOBILE",
    optimization="EDGE"
)

Once deployed, field inspectors use the Maximo Mobile app to take photos of assets during inspections. The Visual Inspection model runs inference on the mobile device, classifying the image and flagging potential defects in real time. If the model detects severe corrosion, for example, it can automatically create a work order for repair and attach the photo as documentation.

The key advantage of Visual Inspection is that it does not require the 18 to 24 months of historical data that Predict needs. A model can be trained and deployed in 4 to 6 weeks, making it the fastest path to AI value in a new Maximo implementation.

Maximo Assistant on Mobile: Natural Language for Field Technicians

Maximo Assistant on Mobile uses natural language AI to help technicians find asset information, review history, and complete work efficiently in the field. A technician can ask questions in natural language and get answers based on the asset's work order history, inspection records, and documentation.

The Assistant is the AI capability with the lowest barrier to entry. It does not require predictive models, sensor data, or labeled images. It requires only that the Maximo asset and work order data is reasonably complete and accurate. For organizations that are just starting their AI journey in Maximo, the Assistant provides immediate value while the data foundation for the more advanced capabilities is being built.

The Maintenance Maturity Curve and Where AI Fits

The AI capabilities in MAS 9.2 map to a maturity curve that describes how organizations evolve their maintenance practices over time. Understanding where your organization sits on this curve is essential for choosing the right AI capabilities to pilot.

Stage 1: Reactive. Fix when it breaks. No AI needed. The organization responds to failures after they occur, with minimal preventive maintenance. Asset data is typically incomplete, and work order history is inconsistent.

Stage 2: Preventive. Fix on a schedule. Still no AI needed. The organization performs scheduled maintenance based on time or usage intervals. The data foundation is being built, with consistent asset hierarchies and work order records.

Stage 3: Condition-Based. Fix when condition indicates a problem. Maximo Monitor provides this capability. The organization has deployed sensors on critical assets and uses threshold-based alerts to trigger maintenance. The data foundation includes consistent meter readings and sensor data.

Stage 4: Predictive. Fix before it fails. Maximo Predict provides this capability. The organization has 18 to 24 months of clean historical data and uses machine learning models to forecast failures. The data foundation includes consistent failure coding, meter history, and work order completion data.

Stage 5: Prescriptive. AI recommends the optimal action. MAS 9.2 Condition Insight provides this capability. The organization has integrated sensor data, historical maintenance data, inspection data, and reliability strategies. The AI interprets all of these data sources to recommend specific actions in natural language.

Most organizations in 2026 are at Stage 2 or Stage 3. The transition from Stage 2 to Stage 3 requires IoT investment and sensor deployment. The transition from Stage 3 to Stage 4 requires data quality investment and data science capability. The transition from Stage 4 to Stage 5 requires the integration of multiple data sources and the watsonx platform that MAS 9.2 provides.

Implementation Path for Organizations Starting in 2026

The recommended sequence for organizations starting their Maximo AI journey in 2026 is a phased approach that builds the data foundation while delivering incremental value at each step.

Month 1 to 2: Audit the data foundation. Look at the asset hierarchy, the meter history, the work order history, and the failure records. Identify the asset classes where the data is complete and consistent enough to support AI. This audit will reveal that 20 to 30 percent of assets have data quality good enough for Predict, while the rest need remediation.

Month 3 to 4: Deploy Maximo Assistant on Mobile. This does not require a strong data foundation and provides immediate value to field technicians. It also builds user familiarity with AI capabilities in the Maximo context, which helps with adoption of the more advanced features later.

Month 5 to 6: Pilot Maximo Predict on one asset class. Choose the asset class with the best data quality from the audit. Train the default models and deploy them. Measure the prediction accuracy and the impact on maintenance planning. This pilot will demonstrate the value of predictive maintenance to executive sponsors and justify further investment.

Month 7 to 8: Deploy Maximo Visual Inspection. Choose two or three high-value inspection types where computer vision can augment human inspection. Train models and deploy them to mobile devices. This does not depend on the historical data that Predict requires, so it can run in parallel.

Month 9 to 12: Pilot Condition Insight. Deploy Condition Insight on the same asset classes where Predict and Monitor are already running. The agentic recommendations layer on top of the underlying predictions, so the data foundation from the Predict pilot supports the Condition Insight pilot. Measure the impact on maintenance decision-making and the quality of the AI-generated recommendations.

Month 12 and beyond: Scale and mature. Expand to additional asset classes, integrate additional data sources, and begin building agentic workflows that coordinate across multiple systems. The MCP Server in MAS 9.2 enables these workflows by allowing AI agents to interact with Maximo data programmatically.

Practical Implications

For reliability engineers and maintenance directors, the maturation of AI in Maximo means that the question is no longer whether to adopt AI, but how fast and in what sequence. The capabilities are production-ready, they are integrated into daily workflows, and they are backed by IBM's watsonx platform.

The practical implication for most organizations is that the bottleneck is not the AI technology. It is the data foundation. Organizations with clean, consistent asset data and well-maintained work order history can deploy Predict and Condition Insight within months. Organizations with poor data quality will spend the first 6 to 12 months on data remediation before AI becomes practical.

The investment in data quality is not optional. It is the prerequisite for every AI capability in MAS 9.2. The organizations that invest in data quality first, pilot AI on a single asset class, and expand from there are the ones that achieve measurable results. The organizations that try to deploy AI across all assets simultaneously without addressing data quality are the ones that produce predictions nobody trusts and recommendations nobody follows.

Bottom Line

The AI capabilities in MAS 9.2 represent the most mature, most integrated AI offering in the enterprise asset management market. Condition Insight, Predict, Monitor, Visual Inspection, and the Mobile Assistant cover the full spectrum from reactive to prescriptive maintenance, and they are woven into the workflows that maintenance teams already use.

The implementation path is clear: audit your data, start with the Mobile Assistant for immediate value, pilot Predict on your best-data asset class, deploy Visual Inspection in parallel, and build up to Condition Insight. The organizations that follow this path will move from reactive maintenance to prescriptive, AI-driven reliability within 12 to 18 months. The technology is ready. The question is whether your data is.