AutoAI
IBM
IBM AutoAI builds trust and clarity by visualizing the entire automated experience—helping users select the best models while cutting work from weeks to minutes.
My role
Product designer
Visual Design Lead
User Experience Designer
Timeline
5 months
March — June 2019
Team
2 Designers
Skills
Research & Strategy
User Experience Design
Product Design & Delivery
Overview
Why should the power of AI belong only to experts?
Many businesses saw the potential of AI but lacked the time, skills, and resources to use it. AI remained locked behind expertise. With AutoAI, we set out to accelerate that journey — making it possible for anyone to build and trust machine learning models.
A new partnership
Research had begun developing the core technology, but it needed a way to connect with people. That’s when Design joined the effort — partnering with Research to shape the work into an experiential solution that users could trust and understand.
Experience principles
From the start, we grounded our work in two guiding experience principles: first, that AI should not require experts to use; and second, that the creation of AI must be accelerated. These principles became the lens through which every design decision was made.
The Challenge
How do we accelerate the creation of AI while closing the gap of trust?
Many organizations sought to harness AI but lacked the expertise, resources, and time to do so effectively. Without trust and transparency, adoption slowed, and tasks that should have been accelerated instead became more burdensome.
Initial user feedback
71
%
Wouldn’t trust an automated AI solution
“
This is a toy. Couldn’t expect it to do better than me.
“
Wouldn’t trust anything ‘auto’ unless there was more transparency.
“
Promising, but I need more information of how the models were generated.
“
It shows metrics, but is missing critical information I would expect.
The Opportunity
What if anyone could build AI models — no experts required?
By simplifying and visualizing the process, we saw a chance to democratize AI and unlock innovation at scale. Each iteration focused on delivering an experience that conveyed transparency, engagement, and delight — making AI not only accessible, but meaningful to use.
Redefined Experience Principles
01
Experts not required
A simplified experience that only requires users to tell us what data to use, what they’d like to predict, and the system will handle the rest.
02
Accelerate AI creation
Build multiple custom AI and machine learning models in minutes or even seconds to eliminate the mundane and time-consuming repetitive tasks.
03
Trustworthy AI
Provide transparency and insight as to what the system did and how the AI was generated, open the black-box of AI automation to build trust.
04
Unique experience
The experience should be equally impressive as the technology, ensure users can see what the system is doing in real time and make it engaging and memorable.
The Solution
AutoAI made AI accessible, transparent, and fast.
We designed AutoAI to guide users step by step — from data prep to model selection — cutting a process that once took months down to minutes.
Efficiency and trust
Accelerated model creation while building user confidence through transparency and education.
Transparent process
Animated, real-time visualizations explained each step — turning a black box into a clear, teachable system.
Simple start
Upload a data source, choose a prediction, and go — AI made accessible from the first step.
Visual insights
Top-down views highlighted shared or unique characteristics, with interactive detail and performance comparisons.
Guided preparation
Combined and cleaned data with intelligent guidance, reducing a process that normally takes hours or days.
Bias detection
Identified outliers early, explained their impact, and provided seamless ways to refine and improve outcomes.
What they are saying now
“
Wow, this is so easy to use and not like the typical product from IBM.
“
This is the simplest experience I’ve used to create AI models. Will start each AI project with this.
“
We achieved better results in AutoAI compared to “manual” ML models.
“
I participated in IBM code cafe today to checkout their AutoAI feature and I must say, WoWwww!!!!!!
“
We were able to speed up modeling iterations from weeks in Python or R down to hours in AutoAI.
“
The updates made are really nice, love seeing the live view of what’s happening.
“
We utilized AutoAI for game rankings that were developed within weeks sparking continuous conversation between fans.
By the numbers
Pushing the envelope has its successes
Results and impact
IBM Data Science Elite
AutoAI has completely changed how the elite team starts any AI based project, using it as their primary tool to jumpstart model creation for all clients.
No. 1 tool
Used for every AI project
Wunderman Thompson
AutoAI helped solve an eight year issue they had been struggling with in only two months, the new AI outperformed previous models by 200%.
200%
Model performance increase
Highmark Health
With the help of AutoAI, they were able to reduce model development to identify patients likely to develop sepsis from 12 months to only six weeks.
90%
Reduction in AI development time
Gartner
AutoAI was one of the major factors that Gartner identified as a differentiator that helped propel IBM to be the Data Science and Machine Learning Platform Leader.
2021 Leader
DS & Machine Learning Platform
