AI Studio

House all-things-AI

Created a unified workspace for Cresta's fragmented AI workflows as the company scaled. This internal platform streamlined processes for the AI Delivery team, enhancing efficiency and model quality. Working with internal users enabled rapid research and iteration cycles as we built a comprehensive system for defining, labeling, training, and evaluating AI models.

Vision

Streamline AI modeling processes across all user personas in a centralized platform.

Problem to be solved:
Lack of unified process or platform for building and managing Cresta's AI models.

Business need:
Accelerate AI modeling to shorten sales cycles.

My role:
End-to-end design leadership from concept to launch.

Outcome:
Reduced labeling time while increasing label completion rates.

Learning:
Even complex, outdated processes can be successfully untangled with persistence.

What happens in AI Modeling?

What happens in AI Modeling?

What happens in AI Modeling?

Who uses Studio?

Who uses Studio?

Who uses Studio?

Problem 1: Fragmented tools

Tasks managed in Google Sheets led to errors, biases, and inefficient transitions between workflows. Sheet limitations hindered diverse labeling types (intent, outcome, entities).

Problem 2: No task management

Communication relied on Slack, creating poor visibility, delayed updates, and ad-hoc tracking solutions that required manual maintenance.

Problem 3: Cryptic Processes

Legacy steps and technical jargon created steep learning curves. Team members memorized procedures without understanding them, which became unsustainable as we scaled.

Research insights

1. Cresta needs its own platform to house all things AI

The AI work was only going to grow as the company grew. Now was the time to spend the effort to build an internal platform.

2. Not everything has to be reinvented

Some of the flows were just messy work that needed to be ironed out in a clearer UX (e.g. task management). But some had to be scratched out and built new (e.g. labeling work).

Solution for problem 1: Unified AI Platform

Studio Platform Launch

Built an all-in-one AI solution with consistency as the guiding principle. Users could tackle various tasks without relearning interfaces, reducing ramp-up time.

Centralized Task Management

Created a hub that served as the starting point for users, critical for managing increased labeling and QA volume.

Studio = Intuitive Experience

Transformed cryptic processes into clear, step-by-step flows. Eliminated memorization requirements and converted complex interfaces (like deployment) into intuitive 4-step processes.

What was the impact?

Increased Output from the AI Delivery Team

The team was able to produce significantly more in less time. This not only freed them from spending countless hours on repetitive tasks but also provided a major boost for launching pilot programs with prospective customers.

Not Just Efficiency—But Delight

The primary goal of Studio was to consolidate everything into one cohesive suite. As we developed the platform, we realized that our internal users were spending most of their day on it, often repeating tasks. To keep the experience engaging, we added small, delightful interactions throughout the user flow. These little touches were well-received and made a big impact on user satisfaction.

Made with lots of 🍕 + 🍷 in Berkeley, CA

Made with lots of 🍕 + 🍷 in Berkeley, CA

Made with lots of 🍕 + 🍷 in Berkeley, CA