K1st BUILD
Build AI models from human knowledge
Company
Aitomatic
Position
Founding Product Designer
Year
2022 – 2023
Market
APAC & US, B2B
Problem
Manufacturing teams lost time and momentum building AI models, stuck with incomplete datasets and heavy prep work.
Solution
Built K1st BUILD, a knowledge-first app that captures expert input to accelerate model development without heavy data reliance.
Impact
•
Cut data prep time by 30%
•
Boosted model accuracy by 15%
•
Improved alignment between engineers and experts by 20%
Problem
Manufacturing companies rely heavily on accurate AI models for predictive maintenance and operational optimization. However, AI engineers frequently got stuck cleaning and preparing data, facing incomplete datasets with limited accuracy. Meanwhile, domain experts who held over 30-40 years of irreplaceable know-how, struggled to share expertise ay large scale.
At Aitomatic, we saw a clear opportunity: what if AI models were built from structured human expertise rather than from imperfect data sets?
Design principles
As Founding Product Designer, I led the design of K1st BUILD: an platform that captures and translates expert input, allowing manufacturing teams to develop accurate AI models faster and with far less data preparation.
With the insights from leadership team, I defined the design approach: structured expert insights could fill data gaps and boost model accuracy while speeding up development.
Design process
I began by approaching with AI engineers and domain experts. Interviewed them, interviewed our internal AI engineers and learned more about the expert-engineer collaboration by studying their JDs.
Interviews and user workflow mappings made it clear where critical knowledge was lost or misunderstood. I iterated on rapid prototypes, collaborating with PMs, Sales, and users to refine each interaction between domain experts’ knowledge-sharing and AI engineers’ model-building flows.
What we built
K1st BUILD, was designed for two key user groups: domain experts and AI engineers.
For the domain experts, they could easily share and refine. They could chat with AI assistants or upload files or even help the AI engineers to validate model quality.
For AI engineers, they could structure the shared expertise into DSL (domain structured knowledge) to build models. From the model pages, they could use data schema generated from the selected knowledge or use any other dataset they have.
I designed to speed up the workflows of each type of user and the collaboration between them, giving a feedback loop. That helped save time in their work and also communications.
Impact
K1st BUILD reshaped AI model-building in manufacturing.
As we measured in a couple of our customers, K1st BUILD reduced data prep time by 30%, freeing engineers to focus on high-value tasks. Model accuracy improved by 15%, shortening timelines of model finetuning from months to weeks.
Even better, the time spent on communications between domain experts and engineers cut by 20%. The new collaborative workflow created a shared language, radically improving how manufacturing teams created value from AI.
