Create your custom ML model
Launched in 2023
Why AutoML works
All the AutoML products out there are complex and time-consuming to build Hive AutoML simplifies this by enabling anyone to create custom models without coding knowledge, all you need is a set of data ready to upload.
For instance, Chatous wants to make sure the platform is safe for users. They will start by collecting the past conversation history data then
Step 1 - upload data to AutoML (AutoML will start training a model to make sure it identifies sensitive text or images)
Step 2 - review model performance (accuracy of the model)
Step 3 - distribute the model from AutoML to Chatous (this can ensure if a user starts sending sensitive text or images, the model will notice and with the support of developers, it would be blurred or sends a notification to the user)
Team
Vivian Y Lead Product Designer
Mei L Product Designer
Richard L Senior Product Manager
Annie C Assistant Product Manager
Sebastian M Engineer Manager
Pam N Senior Frontend Manager
Soumil U Frontend Engineer
Lucy L Account Executive
Role
I lead the product design in 2023 from an idea to now a launched product that’s on version 30. I oversaw the design process from stating the goal of the product to implementation while working with cross functional teams (dev, PM, EM and sales team).
In this portfolio, I will focus on the biggest flow change since the product launch in 2023 - provide user the quickest and clearest way to create an AutoML model.
Shifting Target User Group
Our goal was to push out the product ASAP to gain feedback from users, while sales team hold hands onboarding users. After launching it for 8 months stakeholders decided to shift the product from a B2B product to a self serve product.
Before Redeisgn
Problems started to surface
Unclear Vocabulary Used
“What’s a Snapshot?”Complex Flow
“Didnt i just do that?”Relying on sale team
“Can we get on a call to onboard our new analyst?”
Retention
We needed to make the changes ASAP in order for any user to understand how to create their own custom model. The goal for this iteration is to boost self-serve users on AutoML.
Growth
After we have resolved all the painpoints users have faced since launch and made sure onboarding was clear and easy for users to understand how to train their custom model, data have shown that we have boosted total growth by 42%
Data
- amount of people reduced trying to ask sales to onboard
- flow painpoints, cut, flow counts, decision counts, time spent
- % new clients
