Drawing Search from 30 Minutes to 5: the Similar Drawing Search System Takeuchi Mfg. Built with WOGO

Takeuchi Mfg. Co., Ltd. is a global manufacturer that develops and produces construction machinery. With outstanding development capability and design and manufacturing know-how, particularly in compact construction machinery, it delivers high-quality products to sites around the world. Decades of development have left the company with an enormous archive of drawings, and making efficient use of that past knowledge in new designs had become an important management issue.

That is what led to the AI-powered similar drawing search system developed jointly with WOGO. A designer simply uploads a drawing PDF, and the system instantly retrieves drawings with similar geometry and structure from a vast internal drawing database – cutting part searches that used to take 30 minutes to a projected five. This article covers the background to the project, how the development unfolded, and where it goes next.

1. The problem that troubled designers: “it must be similar to something, but I cannot find it”

In Takeuchi Mfg.’s design work, there was one area where inefficiency had accumulated for years: reviewing existing drawings for reuse, and searching for similar parts.

Decades of development have left the company with an enormous volume of design drawings held internally. When designing a new part, referring to similar past designs is an important step that leads directly to better quality and lower cost. In practice, though, it depended on an individual designer’s experience and memory, or on searching through folders by hand – and situations where something similar must exist but cannot be found, or where simply confirming whether a similar design exists takes a long time, arose every day.

In the production control department in particular, selecting suppliers to quote for new parts involved finding and comparing similar parts, and that took around 30 minutes per part on average. With several such cases arising every day, it was difficult to secure time for the new design and higher-value work that should have been the focus.

This was not a problem for veteran designers alone. For younger engineers, even knowing what past assets existed was difficult, and an organisational solution was needed from the point of view of passing on tacit knowledge.

2. Why WOGO – the promise of letting AI read the drawings themselves

As it considered automating the search for similar drawings, Takeuchi Mfg. was drawn to the approach WOGO proposed. Many existing solutions depend on manual pre-classification or on attaching keyword tags, and their running costs grow as the number of drawings increases.

WOGO proposed something different: having AI analyse the drawings themselves and judge similarity of geometry and structure automatically. By removing the dependence on keywords and processing the drawing data directly, the approach looked ideal for a company holding as vast a drawing archive as Takeuchi Mfg.

Beyond the technical advantages, WOGO’s approach to development also weighed in the decision. The company valued “proposals grounded in our actual workflow, a flexible development stance, and close communication from the PoC stage onward.” Rather than adopting an off-the-shelf package, building the system from scratch around its own operations, together, was exactly what Takeuchi Mfg. had been looking for.

3. The solution – a similar drawing search system running on-premises

The joint development with WOGO produced an AI-powered similar drawing search system. Because drawings are highly confidential information and a source of competitive strength, the system is built and operated entirely within Takeuchi Mfg.’s internal network, on-premises, so it can be used with confidence without drawing data ever leaving the company.

Using it is extremely simple. A designer uploads a drawing PDF or specifies a part number, and the AI analyses the entire internal drawing database, automatically extracting and listing drawings with similar geometry and structure. Work that once meant searching folders by hand, guided by memory, is now finished in a few clicks.

The system does more than search by geometric similarity. It also offers filtering by criteria such as weight and price, so it can be used for cost studies as well as for reuse studies. Designers can move seamlessly from finding a similar part to comparing costs and judging whether it can be reused.

At the heart of the system is technology that has AI automatically vectorise the geometric and structural features of a drawing and evaluate similarity numerically. By having AI judge quantitatively the resemblance a person perceives by eye, the system reproduces a search process that used to depend on the experience of veteran designers.

4. On the project itself – “we could always picture the finished product”

Takeuchi Mfg. also rates the development process with WOGO highly. What it valued in particular was a style of development that repeatedly verified the system against real drawing data from the earliest stages.

“From early in development we received frequent demos and improvement proposals. Repeatedly verifying against real drawing data and refining the features to match how we actually work made it easy to picture the finished product,” the project lead says.

In system development it is not unusual for the user side to be left in the dark about what is being built until it is finished. On this project, a cycle of frequent demos with real data and rapid feedback meant that Takeuchi Mfg.’s team always had a tangible sense of progress.

That co-creative style also contributed greatly to how well the finished system fits day-to-day work. Refining features around how the work is actually done proved essential to producing a system genuinely suited to the company’s operations – something an off-the-shelf package struggles to deliver. Further improvements to the UI and UX are anticipated next.

5. Results today and impact ahead – search time from 30 minutes to 5, and knowledge used across the organisation

The system is beginning to produce effects at Takeuchi Mfg. in both quantitative and qualitative terms.

Quantitatively, there has been a marked improvement in supplier selection for quoting new parts in the production control department. Searching for and comparing similar parts, which used to take around 30 minutes per part, is projected to fall to about five minutes with the system – a reduction of roughly 83% in working time. On a floor handling several quotations a day, that efficiency is expected to add up to a considerable saving over the course of a year.

The qualitative changes matter too. Team members report that “the psychological barrier to searching for past drawings has come down” and that “being able to check similar geometry quickly should speed up design studies.”

There is also real significance for knowledge transfer across the organisation. As one comment puts it, “we are building an environment where even younger engineers can make use of past assets.” The system is beginning to serve as a bridge that carries the tacit knowledge of veteran designers to newer ones. Similar cases that a veteran simply knows from experience are surfaced by the system, so younger designers too are gaining equal access to a vast archive of intellectual assets.

6. Looking ahead – from a search tool to a design decision support system

The next stage Takeuchi Mfg. envisages goes beyond a search tool. Today the system works as something designers actively query for similar drawings; the goal now is to embed it deeply in the design workflow itself and develop it into a mechanism that supports design decisions. Possible extensions under consideration include presenting similar cases in real time for the part being designed, and recommending the best reuse candidates from a combination of criteria such as cost, weight and manufacturing difficulty.

Of WOGO, the project lead says: “we continue to look for proposals that create new value with AI technology, and for improvements grounded in the realities of our work.” The partnership between Takeuchi Mfg. and WOGO is positioned not as a one-off system development but as a long-term collaboration aimed at continuous improvement and new value.

Takeuchi Mfg. also believes this work will be of wider interest to manufacturers facing similar challenges. For companies that want to advance AI-driven DX – especially those that want to trial and build quickly, that want a system fitted to their own operations rather than an off-the-shelf package, and that prefer to improve as they go – working with WOGO can be an effective option.

In manufacturing, a drawing is more than design information: it is an intellectual asset of the company. The similar drawing search system developed jointly by Takeuchi Mfg. and WOGO is at the forefront of the effort to unearth those buried assets with modern AI and raise productivity and quality on the design floor. What the shift from 30 minutes to five really shows is not simply better working efficiency, but a firm step towards an environment in which designers can concentrate on the creative work that should be theirs.