Toyota’s O-Beya Provides Best Practices for Integrating Gen AI to Make Knowledge Management Collaborative and Interactive

The pressure on automobile manufacturers to develop new and improved designs is never ending. Toyota has created O-Beya, which is a Generative Artificial Intelligence (Gen AI)-based knowledge management tool so that designers and engineers can retrieve the latest information to support their work, including understanding how changes in regulations affect new vehicle designs, insight into the latest thinking on the design of powertrains, or the ability to collaborate with other designers to solve an issue. The Gen AI tools in O-Beya scan the digitized content that Toyota has accumulated over the decades, as well as validated submissions from individual staff members, especially those that will be retiring in the coming years.

Currently, O-Beya has over 800 users. The name O-Beya comes from a Japanese term for big room and reinforces Toyota’s collaboration strategy for the disparate design teams to work alongside one another in a big room to accelerate the successful launch of new vehicles.

The objective for O-Beya is to be more than an information repository. From the information contained in O-Beya, teams at Toyota are creating Artificial Intelligence (AI) agents for improving the understanding of fuel consumption and analyzing vibrations, as well as having agents interact with one another to provide recommendations for best practices, such as improving the performance of the vehicle. To be more than an information repository, Toyota works with Microsoft Azure’s OpenAI service and incorporates GPT-4 Retrieval-Augmented Generation (RAG) techniques into the conversational interactions that designers and engineers have when using O-Beya. O-Beya doesn’t replace the need for designers and engineers to build and maintain their knowledge levels, but rather acts as a resource for guidance and support when unfamiliar situations occur.

Toyota has staff taking responsibility for adding sources and ensuring that the content is up-to-date. Furthermore, Toyota has staff that work on validating the responses provided by the AI agents, keeping the human on the loop, so that agents’ responses are logical and practical.

O-Beya is part of the work delivered by Toyota’s Production Digital Transformation Office (PDTO), which is charged with improving efficiency and the workforce by using data and AI effectively.

The long-term objectives for O-Beya are to improve innovation processes at Toyota to support the company’s challenges in electrifying its models, incorporating driver assistance technologies, and competing with the Chinese Electric Vehicle (EV) producers. As with many initiatives of this type, the change management aspects to improve design and engineering processes are as important to the success of O-Beya as the innovative tools that are contained in the solution.

Engineers and designers will be able to retrieve information quicker than in the past, but also via the agents designed to support individual workflows, they can receive insights collected from experts across the company. The change management aspect required includes a culture of open learning and knowledge sharing for workarounds to problems and avoids mistakes being repeated across the company. In addition, by consolidating the institutional knowledge, O-Beya can be a tool to break down silos between teams.

To achieve these objectives requires more than individuals submitting information on an ad hoc or voluntary basis. ABI Research’s Handling the Baby Boomer Exodus in Manufacturing report (AN-6055) outlined how manufacturers need to consider, for example, ways to capture knowledge (via interviews, mentoring sessions, workshops, and storytelling) to avoid tribal knowledge being lost when individuals retire. In addition, the use of individuals that have been identified as experts will often need to change from answering queries in a reactive manner to embracing Gen AI and validating the mechanics of how the models produce the results. As well as validating the output produced, project leads must invest time and resources so that the models are utilizing the most relevant and recent content, also confirming that there is consistency across the company in how the data are collected, labeled, and organized. Toyota has an Operation Management Consulting Division (OMCD) that produces materials to ensure that content that is incorporated into O-Beya and aligns to both industry standards and the company’s expectations. Another culture shift may be that users need to apply critical thinking to the results produced to avoid misinterpreting the content produced or not adding further prompts to fully appreciate how the answers were generated; additional training on effectively using Gen AI tools will be required.

Manufacturers risk falling behind if they don’t invest time in understanding why they need to incorporate AI into their operations and perhaps, more importantly, how they will do so. O-Beya is the result of Toyota considering these issues and is in conjunction with the company’s Global AI Accelerator (GAIA) that was launched earlier this year and looks to incorporate AI into the engineering of the company’s vehicles and on the manufacturing line.