Surendra Katta

What IBM Bob can actually do against Maximo today: the MAS 9.2 native MCP server, IBM's two Maximo Bob skills, MAF Local Dev Mode, a community evaluation, a task-fit table, and where TMG's Max_Interfaces, Max_mcp, and Max_autoscripts fit.

A dated, evidence-tiered reference to the Maximo Assistant and the Maximo AI Service as of October 2026. It traces the path from the 9.0 AI broker through 9.1, the January and April 2026 Feature Channel drops, 9.2 GA's agentic Assistant and MCP server, and the AI Service 9.2.3 patch level. It also covers the security bulletins, a full capability matrix, the model bill of materials, what the Assistant still cannot do, the Assist-versus-Assistant naming correction, and a checklist for verifying each capability in your own instance.

Storekeepers and buyers want to ask the Maximo Assistant to create an item, raise a purchase requisition and check whether a part already exists. This evidence-tiered guide separates what IBM documents today (the Assistant can query supply-chain objects once an admin enables them; similarity is documented for work orders, records and tickets), what IBM has only announced, and what you can build yourself on MAS 9.2 as custom MCP tools — items that land in PENDING, PRs that land in WAPPR and go through your normal approval workflow, and a duplicate-item check — with code, governance and a rollout plan.

IBM framed MAS 9.2 around asset-first AI and agentic workflows. This is an evidence-tiered assessment of what that actually means — the agentic Maximo Assistant, the Maximo MCP server, AI-generated insights with confidence scores, RAG lease extraction, watsonx Orchestrate in Maximo IT, and AI Service AppPoint tracking — separating what product documentation confirms from what only the announcement claims.

A structured reality check on IBM Maximo Application Suite 9.2. Eight gates that decide whether you can and should move, five verdicts mapped to real customer profiles, and an honest accounting of what the release does and doesn't give you.

A complete, sourced walkthrough of Maximo Manage 9.2 — the new Tickets and Alerts applications, AI-generated alert insights, Collaborate moving inside Manage, automated calibration accuracy validation for ISO/IEC 17025 and ISO 15189, six new time zone processing rules, Java 25, and everything new in Maximo Optimizer 9.2.

A seven-part, primary-source series on IBM Maximo Application Suite 9.2. What actually shipped on 25 June 2026, how it compares to 9.1 and 9.0, what changed inside Manage and across the suite, what the AI layer really contains, the upgrade items that bite hardest, and an honest verdict on who qualifies.

Monitor 9.2 drops its IoT Platform dependency and becomes stand-alone. MVI removes Caffe and Darknet and retroactively drops NVIDIA Pascal GPUs. Health, Predict, Maximo IT, Real Estate and Facilities, HSE, Oil & Gas and AIP all shipped real changes. Here's the product-by-product breakdown, and which ones force work on upgrade.

A ranked, sourced watch-list for the MAS 9.2 upgrade — the Db2 12 licence trap that takes production down 30 days later, the removed user management APIs, add-on parity, SCIM migration, two irreversible settings, Java 25, MVI's retroactive GPU removal, and the small ones that still cost you a weekend.

A dimension-by-dimension comparison of Maximo Application Suite 9.0, 9.1 and 9.2 — platform and navigation, AI, applications, Java and database prerequisites, and support lifecycle — built only from IBM's own pages, with every claim we could not verify flagged as such.

A plain-language guide to MAS 9.2 for business and functional leaders: what was released, what the Feature Channel means, how support works, what 8.x customers must check, and how to confirm platform requirements.

MAS 9 gives you four different ways to get numbers out of Maximo — BIRT, the Operational Dashboard, Cognos Analytics, and the self-service BI tier beyond it. Here's how each one actually works, where its ceiling is, and a decision framework for picking the right one for every report you own.

Work Centers with no migration tool, custom Java that won't compile on 17, RMI integrations that die in a container, and an admin mode window that blocks every user. Here's the honest list of what actually breaks on a Maximo 7.6-to-MAS-9 upgrade, and how upgrade teams have prepared for each one.

A six-part series on IBM's own answer to the MAS 9 native-analytics ceiling: watsonx.data's open Iceberg lakehouse, fit-for-purpose query engines, MIF/Kafka extraction, medallion architecture, watsonx.ai and RAG, closed-loop write-back, and an honest head-to-head with Databricks.

MAS 9 native analytics has a real ceiling. This post names it, then makes the honest IBM-native case for watsonx.data — open Apache Iceberg, fit-for-purpose engines, and a bridge to watsonx.ai your MAS AI Service license already opened.

The three IBM-sanctioned ways to move Maximo data into watsonx.data — MIF REST/JSON over OSLC, the Kafka source connector, and MAS 9.1's asynchronous bulk export — plus the Cloud Pak for Data fabric and the SaaS constraint that decides which ones you can actually use.

How Apache Iceberg's ACID transactions, schema evolution, and time-travel snapshots map onto a Bronze/Silver/Gold medallion for Maximo EAM data — with the full Db2 CREATE DATALAKE TABLE syntax, named conformed entities, and the registration mechanics that keep it all in sync.

How watsonx.data routes interactive SQL, ETL/ML, and high-concurrency BI to different engines — Presto, Presto C++/Velox, Spark, Db2 Warehouse, Netezza — over a single Iceberg copy, with the coordinator/worker architecture, IBM's price/performance claims vs Photon, and the RU cost model.

How the Maximo Iceberg lakehouse stops being a reporting layer: custom PdM/RUL models on watsonx.ai, AutoAI for RAG, the Milvus/OpenSearch/OpenRAG stack over maintenance manuals and work-order text, Granite and Llama model routing, and three concrete ways to write AI outputs back into Maximo.

The series finale: a caveated head-to-head — Iceberg vs. Delta, Presto C++/Velox vs. Photon, IBM Knowledge Catalog vs. Unity Catalog, Milvus/OpenRAG vs. Databricks vector search — plus how IKC and Apache Ranger govern Maximo-derived data, LABTRANS PII masking, and watsonx.governance lineage.

How Unity Catalog's row filters, column masks, and lineage graph turn a Maximo-derived Databricks lakehouse into an auditable asset — mapped directly to the Maximo security groups, site/org restrictions, and data restrictions administrators already configure.

MAS 9 ships real native analytics — Operational Dashboards, Cognos, Health, Predict, Monitor — but every one of them has a ceiling. Part 1 names those ceilings precisely and explains why a lakehouse, not another dashboard, is what sits past them.

Four real ways to extract MAS 9 data for a Databricks lakehouse — REST/JSON API, MIF over Kafka and the Data Export framework, DB-direct, and Lakehouse Federation — with real object structures, Kafka broker config, and honest CDC framing.

A table-by-table medallion architecture for MAS 9 data — cleansing rules for WORKORDER, ASSET, MATUSETRANS, PM, and meter data, conformed dimensions, and gold marts for reliability and cost analytics.

Worked Databricks SQL for the five gold-layer questions a maintenance leader actually asks: MTBF/MTTR trend health, maintenance cost rollups, slow-moving inventory, work order backlog aging, and PM compliance — built on Part 3's silver and gold tables.

A scored decision framework for choosing between Maximo Predict's prebuilt models and a custom AutoML/MLflow model in Databricks — data needs, in-house ML skills, AppPoints economics, and how to write a score back into Maximo.

A six-part series on where MAS 9 native analytics stops and Databricks starts — data extraction patterns, the medallion architecture for Maximo objects, five concrete analytics use cases, custom ML versus Maximo Predict, and lakehouse governance.

The Maximo Assistant is not a checkbox — it's a deployment project. The real path: the Maximo AI Service operator on OpenShift, the watsonx.ai connection, training recommendation models on your own data, and the feedback loop that keeps them improving. Mapped to IBM's ten team-exploration tasks and the realistic 68–128 hour pilot estimate.

Generative AI is a governance conversation before it is a technology one. The data-residency decision baked into SaaS-versus-on-premises watsonx.ai, the human-in-the-loop controls (confidence scores and accept/modify/reject) that keep the AI a recommender not a decider, the feedback loop as a governed data process, and the AppPoints licensing that gates who gets access — with a go-live governance checklist.

When something is broken and the technician doesn't know why, this is where the Maximo Assistant earns its keep. How step-by-step troubleshooting guidance walks from symptom to likely cause, how AI failure-code identification recommends failure class, problem, cause, and remedy codes, how Visual Inspection folds in visual diagnosis, and how PM optimization recommends which preventive work to cut, tighten, or extend.

What the Maximo Assistant actually is, the IBM watsonx.ai foundation models that power it, and the Maximo AI Service that connects the two. A grounded introduction to generative AI in MAS 9 — including an honest map of what it does and, just as importantly, what it does not do.

The feature people mean when they say 'AI in Maximo.' How the Maximo Assistant drafts a work order from a plain-English description, replaces filter queries with natural-language asset search, and recommends field values — priority, work type, failure codes, craft — each with a confidence score you can accept, modify, or reject.

A grounded eight-part series on the watsonx-powered Maximo Assistant and the AI features of MAS 9.1 and 9.2 — the watsonx.ai foundation, natural-language work guidance, knowledge capture, guided troubleshooting, the AI Service deployment path, governance and AppPoints, supply chain (querying items, inventory, and PRs natively vs building item-creation, PR-drafting, and similarity tools on the 9.2 MCP server), and an October 2026 capability map. Written for teams who want to know what generative AI in Maximo actually does, and what it does not.

Your best fixes live in your veterans' heads and in long-description fields nobody reads. Here's how the Maximo Assistant surfaces them — knowledge base search across manuals, SOPs, and repair narratives; similar-record detection that transfers knowledge from experienced staff to new staff; and where remote-expert connection fits — which is Maximo Collaborate (the former Maximo Assist add-on), not the generative Assistant.

How Maximo Civil Infrastructure uses Visual Inspection to analyze drone and camera imagery for automated defect detection on bridge decks, piers, and beams; the difference between classic CNN models and the Large Vision Models introduced in MAS 9.1 that are trained for civil infrastructure and need minimal training data; the GPU, camera, and training-data requirements; how visual evidence links to element-level inspection records; and the honest watsonx.ai/FedRAMP reality that determines whether the LVM capability is even available to a government agency.

The heart of Maximo Civil Infrastructure: FHWA-compliant bridge inspection. How the 0-9 NBI component ratings for deck, superstructure, and substructure coexist with AASHTO element condition states; how condition-state quantity tracking works; how the two-year inspection cycle and special cycles (fracture-critical, underwater) are scheduled; how load rating and scour-critical status are tracked; and how NBI-format export supports the federal submission — walked through a full bridge inspection with sample values.

The finale: the full capability-to-regulation crosswalk for Maximo Civil Infrastructure across FHWA NBIS, AASHTO, 23 CFR 650, NTIS, MAP-21/FAST Act, and GASB 34; how transit's FTA State of Good Repair and Transit Asset Management (TAM) plan data are supported; the reporting the application produces for federal submission and management; the deployment prerequisites and AppPoints posture in one place; and a rollout checklist that mirrors a real pilot from portfolio assessment to first federal-format report.

Beyond bridges: how Maximo Civil Infrastructure handles road condition (PCI scoring, distress identification, IRI ride quality, treatment recommendation), tunnel inspection under NTIS with fire/life-safety element tracking, deficiency tracking that prioritizes by severity and safety impact and links to work orders, and condition-based maintenance that triggers work when a rating drops below threshold with Maximo Health supplying the condition score — walked through a full pavement-section example.

What Maximo Civil Infrastructure is, who it is for, and how it models bridges, roads, and tunnels as element-based and linear assets that the standard Manage data model was never built to hold. Covers the AASHTO element taxonomy mapped onto a Maximo asset hierarchy, where the application sits relative to Manage, Health, and Visual Inspection, the honest licensing and deployment posture, and a fully worked bridge-modeling example.

A five-part, infrastructure-specific guide to Maximo Civil Infrastructure as it ships inside MAS Manage: the linear and element-based asset model for bridges, roads, and tunnels; FHWA NBI bridge inspection and AASHTO element condition assessment; pavement (PCI/IRI) and tunnel (NTIS) condition; AI-augmented inspection with Visual Inspection and Large Vision Models; and the full regulatory crosswalk (23 CFR 650, MAP-21/FAST Act, GASB 34, FTA State of Good Repair) with a government-deployment reality check.

With the spine in place, you attach the maintenance decisions. Job plans define the task, PMs define the schedule, asset templates roll both out at fleet scale, and the dedicated Reliability Strategies application brings RCM/FMEA content natively into Maximo — with a vendor library of 800+ asset types and 58,000+ failure modes.

Asset Performance Management sits on top of the Manage spine, not in place of it. Maximo Health scores criticality and risk, Monitor answers 'is it abnormal now?', Predict answers 'when will it fail?', and Asset Investment Planning turns risk into capital timing — and all of them push work orders back into Manage to close the loop.

Before you touch a Maximo application, get the method right. This is Reliability Centered Maintenance reduced to SAE JA1011's seven questions, FMEA/FMECA with the RPN screen, and the decision logic that maps every failure mode to exactly one task type — run-to-failure, PM, condition-based, predictive, failure-finding, or redesign.

The metrics are the feedback loop that tunes the strategy. This is MTBF, MTTF, MTTR, availability, failure rate and OEE from first principles, the bathtub curve explained, and the Nowlan & Heap finding turned into a concrete PM-optimization method you can run against your own Maximo failure history.

Everything in this series assembles into a phased program: Phase 0 legacy cleanse during the upgrade window, Phase 1 the reference-data spine, Phase 2 the maintenance decisions, Phase 3 the Reliability Strategies layer, Phase 4 APM, and Phase 5 closing the loop — the phase that never ends. Each phase has exit criteria so you know when you can move on.

This is where reliability implementation succeeds or stalls. The strategy is only as good as the data underneath it, and the data has a hard dependency order. Here are the five realistic load tools, the eleven-step load sequence the MBO layer forces on you, per-object load specifics including the FAILURELIST restricted-object gotcha, and the validation and cleansing that must happen before you load anything.

Methodology becomes reference data here. The reliability spine is four stock MAS 9 Manage applications — Failure Codes, asset criticality, Condition Monitoring, and Meters — that together turn free-text guesswork into a controlled vocabulary, risk-based effort, and automatic condition-based work order generation.

An end-to-end, seven-part program for standing up a real reliability strategy in IBM Maximo Application Suite 9 — from RCM and FMEA fundamentals, through the Manage reliability spine, the Reliability Strategies app, and the APM layer, to dependency-correct data loads and a phased, upgrade-window rollout.

A practitioner's deep dive into the generative AI built directly into Maximo Manage — Work Order Intelligence (9.0), the Maximo Assistant (9.1, expanded in 9.2), and the AI Configuration application — and how it differs from the standalone Health, Monitor, Predict, and Visual Inspection suites.

A practitioner's deep dive into asset and location management in Maximo Manage 9 — asset and location hierarchies, rotating items, Asset Configuration Manager for complex serialized assets, linear assets with GIS, and meters with condition monitoring.

The practitioner's guide to configuring and administering Maximo Manage 9 — Automation Scripts, Workflow Designer, Application Designer, Database Configuration, domains, security, Operational Dashboard, new tools like Formulas and in-MAS Application Configuration, and the admin split between MAS Suite and Manage.

A practitioner's deep dive into the Maximo Manage 9 scheduling stack — Gantt-based Graphical Scheduling for outages and projects, Graphical Assignment for short-horizon dispatch, the crafts/crews/calendars foundation, and the AI-driven Field Service Management solution introduced in 9.0 with its Scheduling and Dispatching dashboards, extended in 9.1.

A practitioner's deep dive into Inspections in Maximo Manage 9 — how the Inspection Forms Work Center became a Maximo Application Framework app, the refreshed Form Builder, question types and conditional logic, scoring and pass/fail, follow-up work orders and meter updates, form versioning, and offline mobile execution.

A practitioner's deep dive into inventory and procurement in Maximo Manage 9 — the ITEM and INVENTORY objects, storerooms, costing, reorder processing, the PR-to-PO-to-receipt-to-invoice flow, contracts and vendors, and the role-based Issues and Transfers app that carries forward key Inventory Work Center workflows.

A practitioner's deep dive into Maximo Mobile for Manage 9 — why it replaced Maximo Anywhere, how its offline-first REST and sync architecture works, the Technician, Inspections, and Inventory apps, app configuration in Application Configuration, RBAC and offline security, and how to migrate off MobileFirst.

A practitioner's deep dive into Maximo Reliability Strategies — how FMEA and RCM connect failure modes to job plans and PMs, the built-in library spanning 800 asset types and 58,000 failure modes, RPN-driven actions, custom strategies, and the AI assists added in Manage 9.1.

A practitioner's deep dive into IBM Maximo Manage on MAS 9 — eleven parts covering the new MAF role-based applications, Operational Dashboard, Reliability Strategies, Inspections, Work Management, Graphical Scheduling, Maximo Mobile, embedded AI, and how to configure and administer it all.

A practitioner's map of how Maximo 7.6 EAM became Maximo Manage in MAS 9 — the move to containers on OpenShift, the Maximo Application Framework, AppPoints licensing, and why your MBOs, scripts, and database survive the journey intact.

Work Centers are removed in Maximo Manage 9.0. Key experiences moved into Maximo Application Framework role-based applications, Maximo Mobile, and Operational Dashboard surfaces. Here's what changed, where the new apps live, and how to move your users across without disruption.

The heart of Maximo Manage — a practitioner's deep dive into the work order lifecycle, job plans with the new Qualifications field and milestone tasks, preventive maintenance and routes, plus the Work Order Planning and Work Queue Manager applications added in Manage 9.

Technology does not fail. Programs fail. Learn the governance, change management, and adoption strategies that separate successful Maximo Predict deployments from expensive shelf-ware.

The complete guide to model development in Maximo Predict. Model types, feature selection, training, validation, quality metrics, and what to do when your first model is terrible.

Your predictive models are only as good as your data. Learn the 5 data types, 5 quality dimensions, and 5 deadly sins that determine Maximo Predict success.

Building the model is half the job. Learn how to deploy, monitor, retrain, and build feedback loops that keep your Maximo Predict models accurate over time.

Step-by-step guide to deploying Maximo Predict in MAS. Entitlements, configuration, data connections, scope definition, and selecting a use case that wins.

Stop guessing which use cases work. Real-world Maximo Predict patterns across 4 industries with specific assets, failure modes, data sources, and measurable results.

Predictions without integration are dashboards nobody uses. Learn the end-to-end data flow from sensors to work orders across MAS applications.

Cut through the hype. Learn what IBM Maximo Predict actually does, how it fits into MAS, and why predictive maintenance is not the crystal ball vendors promise.

The definitive 8-part series on IBM Maximo Predict. From data foundations to enterprise governance, everything you need to deploy AI-powered predictive maintenance.

Model lifecycle management, retraining schedules, quality governance, scaling across sites, and the organizational change that makes or breaks enterprise visual inspection programs.

The complete MVI model-building workflow. Data collection strategies, image labeling best practices, training configuration, accuracy metrics interpretation, and iterative improvement.

How computer vision works for asset inspection without the PhD. Classification, detection, segmentation explained. No-code model building in MVI demystified.

Get your MVI model out of the lab and into the field. Inference pipelines, confidence threshold tuning, false positive management, A/B testing, and production monitoring.

Everything you need to plan MVI infrastructure. Five deployment options (SaaS, on-prem, Azure, client-managed), NVIDIA GPU requirements, storage sizing, and AppPoint licensing explained.

Deploy MVI Edge for real-time inference at the camera location. MQTT alert pipelines, edge diagnostics, hardware selection, drone integration patterns, and field deployment strategies.

Complete MVI installation walkthrough. Prerequisites checklist, MAS 9.0/9.1 changes, operator deployment, GPU configuration gotchas, verification, and your first 30-minute project.

Connect MVI to Maximo Manage for automated work orders, Health for condition scoring, Monitor for visual alerts, and Predict for failure enrichment. The closed-loop inspection architecture.

What IBM Maximo Visual Inspection is, why computer vision is transforming asset inspection, and how MVI fits into the MAS ecosystem. Real use cases with real numbers.

Deploy MVI to the field with MVI Mobile on iOS/iPadOS. Core ML export, on-device inference, device requirements, offline workflow, and mobile deployment best practices.

Complete REST API reference for IBM Maximo Visual Inspection. Authentication, endpoints, code examples, automation patterns, and the vision-tools CLI.

The definitive 12-part series on IBM Maximo Visual Inspection. From computer vision fundamentals and model building to mobile, edge, and drone deployment, enterprise governance, REST API reference, and troubleshooting.

The definitive troubleshooting guide for IBM Maximo Visual Inspection. GPU configuration, training failures, deployment issues, mobile and edge problems, integration errors, and answers to every FAQ.