Accelerate AI Model Development with Expertise
Challenges
Manufacturing companies struggled to build AI models quickly due to incomplete and noisy datasets, requiring significant time for data preparation and delaying progress in high-stakes use cases.
Solution
We introduced K1st BUILD — a knowledge-first approach, enabling AI model building workflows to use insights from experts, reducing reliance on datasets while speeding up model development.
Reduced preparation time by 30%
Improved model accuracy by 10–15%
Enhanced team alignment by ~20%
Company
Aitomatic
Position
Founding Designer
Year
2022 – 2023
Market
US and Asia, B2B
Challenges
Imagine trying to predict when your fridge will stop working, but you only know a few cases where fridges have broken—and each story is incomplete. One person mentions strange noises, another recalls it stopped cooling suddenly, but no one knows the full picture. You’d spend hours guessing what went wrong and still have no clear solution.
This is the challenge manufacturing companies face when building AI models for predictive maintenance. Unlike industries with abundant, well-organized data, manufacturing environments often have incomplete and messy datasets. Engineers are forced to fill in the gaps, piecing together scattered information to create models.
Waiting for more failures isn’t an option when the stakes include costly downtime and millions in losses. Without a better way to handle these gaps, engineers spend far too much time preparing data instead of developing reliable models.
Solution
Now imagine you had an expert technician standing beside you while troubleshooting your fridge. They can tell you that the strange noise is caused by a clogged fan or that the sudden warming indicates a coolant leak. Instead of waiting for more fridges to break down, their knowledge helps you identify and solve the problem faster.
This is exactly what K1st BUILD does for manufacturing AI. The platform enables domain experts to share their knowledge in a structured way, turning it into a blueprint engineers can use to fill gaps in incomplete datasets. By combining expertise with AI workflows, K1st BUILD accelerates model development without relying on perfect data.
As the Founding Designer, I collaborated with CNA (China & Northeast Asia Company by Panasonic Group) and Aitomatic’s engineering teams to create a platform that transforms how experts and engineers work together:
Experts can document and structure their insights into reusable components, which engineers integrate directly into AI models.
Engineers have tools to adjust schemas and refine models dynamically, ensuring precision even with inconsistent data.
Feedback loops allow engineers and experts to validate and improve the models together, creating a continuous improvement cycle.
During testing, users highlighted the need for seamless transitions between knowledge input and model development. In response, I refined the platform to ensure that expert insights flowed effortlessly into AI workflows, creating a unified and intuitive experience.
Impact
K1st BUILD transformed the way AI models are created by prioritizing simplicity and collaboration. Engineers and experts now work together seamlessly, building better models in less time.
By integrating domain expertise directly into the process, the platform reduced data preparation time by 30%, enabling engineers to focus on core model development. In predictive maintenance, model accuracy improved by 10–15%, achieving results in weeks that once took months.
K1st BUILD also strengthened teamwork, with engineers and experts reporting a 20% improvement in alignment, leading to more productive and impactful collaboration.
