Minerva Studio @GEP
A no-code platform designed to help organizations train, deploy, and manage AI/ML models with ease.
At GEP Worldwide, I joined the team to design workflows and features for an enterprise AI/ML platform. My focus was on making model training and orchestration easier for non-technical users, while still giving data scientists the flexibility they needed. Working closely with a senior designer, product managers, and engineers, I learned how to navigate the challenges of enterprise UX and data-heavy workflows.
Due to NDA restrictions, only high-level details are shared here.
Problem
How might we help non-technical users train and run models without code, while giving data scientists flexibility to customize?
Outcome
Non-technical users could train and run models without coding, while experts still had the flexibility to go deeper. This opened AI to a wider set of teams and made adoption across the organization much smoother.
Uncovering issues
findings & insights
The team conducted thorough research to understand how users interacted with existing AI platforms and where they faced challenges.
01
Lack of guidance
Beginners often felt lost after logging in, with little direction on where to begin or how to move through the workflows.
02
Need for no-code options
Business users expressed a strong preference for interfaces that let them upload data and see results without writing code or understanding complex algorithms.
03
Overly complex tools
Non-technical users struggled to apply AI models because current platforms assumed advanced expertise, creating a steep barrier to adoption.
Product Vision
Create a no-code platform that makes building, training, and deploying AI models possible without specialized skills, opening AI to business users while still serving experts.
Initial Concepts
ideas & flows
We explored different ways the platform could introduce workflows, models, and orchestration in a simple, guided manner. The goal at this stage was to sketch broad directions for layout and navigation before moving into detailed design.
High fidelity
prototype
To bring the concept closer to reality, we built a high-fidelity prototype in Premiere Pro and Figma, creating first-person videos to simulate the eyeband experience.
Home Page
The home page was designed to give users a clear starting point after login. It surfaces active workflows, provides an overview of available flows, and links to supporting resources, helping beginners understand where to start while giving returning users quick access to their work.
Models Page
This page introduced clear pathways for using models — either starting with a pre-trained one or creating a new one. By separating categories and simplifying their descriptions, we helped users understand options quickly and choose the right path without being overwhelmed by technical terms.
Orchestration Page
This page was designed as a canvas where users could build and customize workflows. An initial structure provided guidance for beginners, while flexible nodes allowed experienced users to rearrange and refine flows.
