IBM has shared details about a recent auto sector project that drew on its Granite series of language models, consulting services and know-how from a major services partner.

The client was the Canadian unit of auto maker Mitsubishi Motors. IBM worked with WongDoody, a creative agency owned by Infosys, to build an AI tool the companies refer to as The Intelligent Companion. It’s notable because the majority of generative AI applications in the enterprise have not yet moved into production. Mitsubishi’s new IBM-developed tool is an early example of a project that has already made that leap, and quite quickly: the companies say they moved from initial discussions to launch in about three months.

IBM, WongDoody and Mitsubishi brought The Intelligent Companion online on March 5. It’s embedded into a set of webpages that Mitsubishi’s Canadian unit has set up for the 2025 Outlander, the auto maker’s flagship SUV. The AI can answer user questions about the vehicle’s technical specifications and interior design.

WongDoody’s contributions to the project included the development of the three-dimensional visuals in the tool’s interface. When customers ask about the 2025 Outlander’s audio system, The Intelligent Companion displays a high-tech diagram of the vehicle that shows the location of the speakers and related sound equipment. Circles overlaid on the diagram mark the placement of each audio device within the SUV. A similar diagram explains the airbag system. Questions about other vehicle features, in turn, bring up a brief natural language response and illustrations.

WongDoody collaborated with IBM’s consulting group on the project. IBM provides a consulting offering called Client Engineering that is designed to help companies quickly build new applications. Additionally, the tech giant has partner-facing professionals who focus on coordinating work with consulting partners such as WongDoody. “The Ecosystem Engineering team works closely with our partners to build AI-powered, fit-for-purpose assets and solutions with IBM technology, including focusing on revolutionizing automotive innovation with AI-driven solutions that redefine the customer journey,” Savio Rodrigues, the vice president of services partners at IBM, told Boardroom Insight.

The IBM technologies that power Mitsubishi’s new AI assistant include a language model series called Granite. The algorithms in the lineup are available under an open-source license, an important detail for many enterprises. Open-source LLMs have gained significant traction over the past year because they are often more cost-efficient than proprietary LLMs. Additionally, they can be deployed on a company’s internally-managed infrastructure, which is important for risk-averse enterprises. 

Granite was trained on data about the Outlander 2025 to equip it with the ability to answer customer questions. According to IBM, the engineering workflow involved two of its AI development tools: IBM watsonx Orchestrate and watsonx.ai. Both products are designed to ease the creation of AI applications, but they do so in different ways. IBM watsonx Orchestrate includes prepackaged building blocks for building machine learning software. The other IBM tool that supported the development effort, watsonx.ai, eases tasks such as customizing AI models for specific tasks.

IBM and WongDoody moved quickly after Mitsubishi entrusted them with the project. According to the companies, their engineers took four weeks to develop the initial proof-of-concept prototype of the AI tool. The production version embedded in Mitsubishi’s Canadian website came online eight weeks later.

In the auto sector, IBM’s AI consulting work is not limited to customer-facing software. “By harnessing real-time generative AI, we seamlessly integrate vehicle telemetry, customer profiles, dealership services, and maintenance data to anticipate and address customer needs proactively,” Rodrigues said. “Our advanced predictive analytics, context-aware automation, and intelligent data orchestration not only enhance user experiences but also drive dealership efficiency. This positions automotive brands at the cutting edge of customer-focused, data-driven transformation.”

Companies are exploring applications for AI in every part of the auto sector, from the dealership ecosystem to car plants. BMW has built a ChatGPT-like tool that helps employees find internal information more quickly. Elsewhere, multiple automakers are studying how generative AI can be used to speed up the vehicle design process.

text

text