Launching our first AI Efforts
After a series of pivots, around 2017, we began providing AI solutions to the market, and we received a wide variety of requests, from analysis of human resources (HR) data, to crop harvest prediction. By far the most common inquiry was about automating inspections using images.
Visual inspection has long been one of the areas where AI can be applied, and many vendors had solutions in the market, however we received interests from firms that were excited about our AI’s strengths, such as the ability to work with small data sets and the ability to train locally, without the need for dedicated hardware (GPU).
While we verified the technical applicability not only to visual inspections, but also to various projects, our first on-site implementation of a AI system was for a visual inspection project. The project aimed to reduce inspection time and manual labour by combining our AI with the imaging equipment the customer already had, and it is still in use several years after its introduction.
I myself originally worked in the IT field and had many years of experience developing software for office use, so I felt great joy in the fact that our AI was operating and being useful on the factory floor. It was the moment when I felt the transformative power of AI.
Exploring Go-To-Market Strategies
Following this we received a lot of inquiries about our visual inspection solution, so we started developing it as our flagship product. At the time we named it SPECTRO, which comes from the Latin meaning “appearance” and also reminds us of the English word “inspection”.
As visual inspection projects typically take a long time from initial data assessment to launch, we proceeded with development based on the concept of making it easy for users to test our AI using a cloud module, and then move models seamlessly for use offline. After launching this service, the number of customers who wanted to try out SPECTRO through cloud trial increased, but on the other hand, the number of on-site implementations did not increase at all – and we sometimes received harsh criticism from customers regarding the product and lack of features.
Simply put, one of the reasons was that SPECTRO could not provide enough performance to be used in the field. Of course, it could have been used by applying pre-processing, tuning, and customizing it to meet the challenges of each individual customer, but we began to search for ways to turn this into a product.
HACARUS Check’s Unique Value Proposition
In the midst of all this, we came to the conclusion that we should rethink the product as an all-in-one package that includes not only software but also hardware, including image capture functions. Due to technological advances in AI performance, it is becoming increasingly rare for a single algorithm to make a huge difference. In fact, it has become clear that the method of image capture has a large effect on inspection performance.
Although there were already many competitors we sought to bring out the strengths of our company’s AI and searched for areas that could not be solved with existing inspection equipment, and finally came up with the current concept for HACARUS Check.
The value proportion of HACARUS Check is to save labor time in inspecting three-dimensional objects through 360-degree inspection that combines a robotic arm with AI.
HACARUS Check has been completely renewed as a native application for Windows while inheriting the algorithm assets developed for SPECTRO. This makes it possible to perform one-stop processing from robot control to AI analysis on a Windows PC, while also making it easier to use as standalone software.
Following this focus, we once again received interest from the market, and when we heard from a customer who deployed HACARUS Check that it solved a problem that they had been working on automating for many years (and had previously given up on because it was too difficult), I was convinced that we were on the right track. The case study for this customer is linked as a video below:
HACARUS Looks to the Future
As the working population in Japan continues to decline, many of our customers in the manufacturing industry are facing the urgent task of reducing labor needs, and increasing automation in factories. Against this backdrop, HACARUS Check has come a long way with many twists and turns, and we would like to continue to evolve it further so that we can help solve many more use cases and issues.
To this end, it is essential for us to have customers try out HACARUS Check in a variety of use cases and give us candid feedback. Please feel free to share your challenges with us. We may not be able to solve them easily, but we will do our best, together with our colleagues who believe in the potential of our AI.
Its been 10 years since our founding, and we will continue to strive to create products and solutions that will help solve problems in the manufacturing field for many more years to come!