K1st BUILD
Build AI models from human knowledge
Company
Aitomatic
Position
Founding Product Designer
Year
2023
Market
APAC, B2B
01
Problem
Manufacturing teams used AI models for predictive maintenance and operational optimization, but usable production data was often incomplete.
AI engineers spent significant time cleaning data, filling gaps, and translating input from domain experts with decades of operational experience. Knowledge was frequently lost or misunderstood during these handoffs, slowing model development and limiting accuracy.
02
Solution
K1st BUILD brought expert knowledge directly into the AI model-building workflow.
Domain experts could share knowledge through guided AI conversations and uploaded documents, then review model quality. AI engineers could translate that input into a domain-specific language, generate data schemas, combine it with available datasets, and refine the model.
The workflow gave both groups a shared way to clarify knowledge gaps, trace how expert input shaped model logic, and validate the result together.
03
my role
As Founding Product Designer, I shaped the product around two connected workflows: experts contributing and validating knowledge, and AI engineers turning that knowledge into model logic and training inputs.
I led research with customer experts and internal AI engineers, mapped where information was lost between them, and defined the knowledge capture, structuring, schema generation, model-building, and validation experiences.
I worked with product, engineering, sales, and customers through ideation and validation.
04
impact
Across early customer engagements, K1st BUILD reduced data preparation time by 30%, improved model accuracy by 15%, and cut coordination time between domain experts and AI engineers by 20%. Some teams, such as Panasonic in China and Lawson in Japan, also reduced model fine-tuning timelines from months to weeks.
