McDermott’s AI assistant helps engineers find and access information quickly

At global engineering and construction firm McDermott, AI is top-of-mind. Soon, employees can choose between not one but several AI applications to automate and optimize engineering processes. For example, the ReCo tool that identifies mistakes in Piping & Instrumentation Diagrams. This article, however, is about yet another development: InfoWiz – an application that uses a Large Language Model (LLM), designed to save time and efforts on engineering projects.

InfoWiz: 3x AI

McDermott's projects always involve large amounts of data and text. Infowiz is intended to help find and access relevant information more quickly and easily.

InfoWiz is comprised of three different modules, each with a specific function and purpose. Behind the scenes, the solution is powered by the Microsoft Azure AI LLM technology, ensuring the data remains secure and private.

Module 1: Interactive Document Query

Maximizing ‘AI as a self-service’, this module enables engineers to access and analyze any documents with ease. Users can upload documents and ask question in natural language: such as “Extract weight and dimensions of the equipment”, “List the applicable standards”, etc. The system will provide relevant answers based on the content of the documents. This module can be used to summarize, extract, enrich and transform documents as needed. Users can quickly get the most out of their documents and gain valuable insights. In addition, The module offers automatic translation from and to specific languages to best serve international projects.

Module 2: Corporate Documents

In this module, a selection of corporate documents has been preloaded in the application. This includes documents from McDermott’s global engineering library, such as standards, best practices, manuals, and guides that are commonly used for project execution. Queries retrieve accurate information from the database, ensuring answers are based on McDermott trusted sources. For instance, "What is the best type of cable to use to connect operating buildings?". The application responds: "[...]Redundant fiber-optic cables [...]" and provides a reference to the document where the answer come from.

The application uses semantic search, which allows users to run queries in their own words, without having to match exactly the original text. This is different from a traditional search engine that depends on keywords and phrases.

McDermott's engineers save time finding the information they need faster and can deliver more value to customers.

Module 3: Virtual Assistant

If you ever used ChatGPT before, you know how this works. The main difference here is that the data processed remains safe and private, with all interactions kept within McDermott.

Users can engage in human-like interactions with the virtual assistant. This can be used for a large variety of tasks such as parsing engineering standards, brainstorming new ideas or fixing a spreadsheet report. Answers to these questions are not based on internal data from McDermott but come from the LLM itself. In all modules, data is never stored, however users have the option to export the questions and answers to their device.

Efficiency across disciplines

Behind every application is a brilliant idea. As an Information and Data Management Specialist with an interest in AI, Maxime Moulin wanted to use the LLM technology to improve processes and break down the silos between different engineering disciplines. Fast forward six months and the idea turned into an application that is currently being used by a select group of digital pioneers at McDermott (read: project managers, department managers, engineers). Moulin: “This is the essence of digital transformation: leverage technology to do things differently and be more efficient”.

Partnership

Moulin praises “the innovative AI team at McDermott and the power of the VIKTOR platform that makes the InfoWiz development possible.” This is proven by the fact that there are already quite a few AI applications developed at McDermott, such as the ReCo tool for patterns recognition, several deep learning and prediction tools, and now applications that use natural language abilities and generative AI, like InfoWiz. “VIKTOR is more than just a platform for agile product development, they are also a partner to McDermott. They support us with our technological needs, and we contribute to their service development. On top of that, we received excellent support and have been brainstorming together, exchanging ideas, and even sharing code. One more key aspect is that VIKTOR is a web-based platform, allowing McDermott’s users to access it easily from our offices globally, whether in Netherlands, India, Malaysia, or the USA.”


What’s next?

After the initial launch, requests to further customize InfoWiz started coming in fast, with myriad of potential use cases. For now, the AI team is planning to add more documents from the corporate management system to the knowledge base. Next, they will investigate integrating with other McDermott’s applications, so that users can access project-specific information.

Share

Related Customer Cases

McDermott leverages AI to save valuable engineering hours with user-friendly VIKTOR apps

Read more
Let us show you how VIKTOR
can help you solve any challenge
Book a demo