Watsonx.ai Use-Case

The system utilizes IBM watsonx.ai’s suite of tools to generate and optimize job plans within IBM Maximo. The system ingests large volumes of unstructured operational data - such as equipment manuals, technician notes, and maintenance records - and use this data to recommend or create detailed, context-aware job plans.  
‍

The core goal is to demonstrate how job plan creation and enhancement in Maximo can be automated using watsonx.ai.
‍

The image sequence below, Figures 1 to 4, demonstrate how the watsonx Assistant can be used to interact with maintenance documentation through a conversational interface. In this example, shown in the sequence of figures below, a user queries the assistant about brake manual instructions. The assistant retrieves relevant information from a document that was previously uploaded to Watson Discovery. This allows the underlying Large Language Models (LLM) to access and understand the manual content when forming its response. Once the assistant provides the instructions, it also prompts the user with an option to automatically generate a job plan based via a Maximo API on the retrieved information.  
‍

‍

High Level Architecture

The implementation integrates IBM watsonx.ai services with the Maximo Application Suite through an intermediary microservice. This microservice facilitates the retrieval of information from the specified data source and automates the creation of job plans based on the LLM’s response.
‍

System Components

1. User Input Via Watsonx Assistant - Users interact with the system through watsonx Assistant, which is designed for dialogue and interaction, asking questions or requesting job plans in natural language.
2. Action Framework - Behind the scenes, these inputs trigger predefined Actions, which orchestrate responses by pulling data from multiple sources, all of which is imported and executed in watsonx Assistant.
3. Watson Discovery - This focuses on searching and retrieving relevant information from documents. It is used to ingest and analyze unstructured documents and extracts key information which makes it queryable by the watsonx Assistant.
4. watsonx.ai Studio - Studio is to deploy LLMs and generate content based on the user prompts and data. When natural language processing or structured generation is required, the information extracted using Watson Discovery is passed to an LLM via watsonx.ai Studio. The options for the LLMs in watsonx.ai are:
   
   1. Granite: IBM's proprietary foundation model, designed for enterprise-grade AI tasks with a focus on trust, transparency, and domain-specific performance.
   2. Llama: Developed by Meta, optimized for efficiency and performance on open research and general-purpose language tasks.
   3. Mistral: Created by Mistral AI, known for being lightweight, high-performing open-weight models optimized for speed and deployment flexibility.
      
      There is an option to import your own model from HuggingFace, an open-source platform for LLMS, as well.
5. Data Extension & Integration - The system includes extensions for Watson Discovery, Watson Assistant, and the Maximo API, enabling seamless communication between components.
6. Microservices for Business Logic - A lightweight microservice layer processes inputs, formats requests, and prepares POST calls to Maximo Manage APIs. This layer acts as a bridge between AI outputs and Maximo’s internal structure.
7. Database Interaction - The microservice can also directly interact with the Manage Database via JDBC if needed to validate or retrieve existing job plan data. However, this implementation was not included in the example shown in the overview.
8. Maximo Manage - Finally, the job plans are either uploaded into Maximo via an API or used to suggest actions within Maximo Manage, where planners can review and approve them.

‍

IBM Cloud Services

All watsonx services can be found on cloud.ibm.com. They can also be hosted on other platforms like AWS, but in this example, everything is hosted on IBM Cloud. From the IBM Cloud dashboard, click on the Navigation Menu as shown in Figure 6 below.
‍

Select the resource list option from the navigation menu as shown in Figure 7 below.
‍

In the Resource list, you will see an AI/Machine Learning dropdown option, as seen in Figure 8 below, which shows you services from IBM such as watson Discovery, watsonx Assistant, and watsonx.ai Studio. These are the tools referenced in the High Level Architecture section above.
‍

Key Takeaways

This use case shows how AI can help simplify and speed up the creation of job plans in Maximo. By using tools from watsonx.ai, the system can reach through large amounts of technical information – like manuals, technician notes, and maintenance records – and turn that into useful, detailed job plans. Instead of searching through documents manually, users can simply ask questions through the AI assistant. It pulls the right information, understands it using AI, and offers to generate a job plan automatically.
‍

Behind the scenes, several powerful tools work together: Watson Discovery helps the assistant understand documents, watsonx.ai Studio generates smart responses using LLMs, and Maximo APIs connect everything to the system. It’s all designed to make the process easier, faster, and more accurate. While this example focuses on understanding maintenance instructions and creating job plans, the same approach could be used in other areas too – making it a great starting point for using AI in day-to-day operations.  

‍

‍

‍

‍

‍

‍

‍