Toyota is using AI to slash 45,000 department-specific terms down to 5,000
Every car that rolls off a Toyota assembly line carries thousands of individual parts. But underneath the hood of the company’s own operations lies a mess of incompatible languages — 45,000 specialized terms scattered across planning, production, marketing, and sales departments, none of them speaking the same dialect. The company now plans to fix that with an AI-powered system that will shrink those 45,000 terms down to 5,000 standardized ones by 2028.
The problem runs deep. A part that the planning department labels “A1AC — no audio configuration” might show up in the production system as code “136S.” The same component can have half a dozen names depending on which vehicle model and which manufacturing process it passes through. Of the 560 steps from the first draft of a spec sheet to the final sale of a finished car, 195 of them — roughly a third — exist solely to translate terminology between departments.
This fragmentation traces back to 1950, when Toyota split production and sales into two separate companies. Each kept its own records. Planning teams could name parts however they wanted on spec sheets. Production used 200-character codes. Marketing used 80 characters. Sales squeezed everything into 15-character abbreviations. By the time the two companies merged back in 1982, everyone had already built their own business systems around those incompatible standards. The mess simply fossilized.
Today, employees across departments collectively spend 310,000 hours per year just translating terms from one internal language to another. The system involves 800 separate business applications, each with its own data conventions.
Toyota has been quietly working on a fix since 2021, when a board member proposed cleaning up the spec management process. The resulting project is called OMUSVI — Organized Master Unified System for Vehicle Information. A dedicated team was assembled the following year.
The AI-powered central system will sit on top of all 800 existing systems and translate between them in real time. It won’t just handle vocabulary — the system will also be connected to Toyota’s annual production planning platform. Once live, the AI will automatically calculate the quantities of each component needed and push that data directly to suppliers. Currently, suppliers get blank annual production forecasts showing only the total number of vehicles planned — no part specifications or quantities — and must spend months figuring out their own capacity requirements. Some suppliers burn 720 hours on that single task.
The OMUSVI team mapped the entire material and information flow inside Toyota’s production network, documenting every handoff between departments, every document exchanged, every approval required. The resulting flowchart covers 560 steps.
Toshinao Wada, the executive leading the project, described the old information architecture as “a Pandora’s box nobody dared to touch.”
In November 2023, the project hit a wall. Departments pushed back, worried that the unified system would eliminate their roles. Wada admitted he thought the team couldn’t resolve those internal conflicts on its own.
The turning point came in March 2024, when then-Toyota CEO Koji Sato visited the project office by chance. He saw the wall-sized flowcharts and became genuinely interested. Sato publicly endorsed the digital unification effort, giving it the top-level backing it needed to push through resistance.
Toyota plans to integrate OMUSVI with the annual production planning system in time for the 2028 fiscal year. The unified term database will also be opened to suppliers, giving them access to the same real-time production data as Toyota’s own teams.
When the system goes live, Wada said, “the efficiency of the entire supply chain will improve, further strengthening Toyota’s overall market competitiveness.”
The question now is whether the world’s largest automaker can untangle 80 years of accumulated internal jargon before the next production cycle forces its hand.