AI Builder

#

Overview

AI Builder is a tool to improve business performance by predicting outcomes and automating processes. The easy point-and-click experience targets users without a coding or data science background and allows them to easily incorporate AI into their business workflow. The growing number of AI Builder templates offer a wide range of solutions for use in applications and automated flows.
View AI Builder

AI Builder build page

The challenge

Democratizing AI for business

Because the target AI Builder users typically do not have machine learning experience, it is crucial to help them feel they can trust and have confidence in this new technology. The entire experience is designed to be simple, clear, and to only reveal AI complexity if necessary. We regularly measured AI Builder through user studies and customer interviews to make sure we were on track with our simplified and transparent approach.

AI Builder chart

My role

As a Senior Designer on the team I helped shape the initial structure and design of AI Builder from the ground up prior to its June 2019 launch. Although I continued to influence a wide number of AI Builder areas, my main focus shifted to simplifying the Prediction template and enhancing the value proposition of AI Builder through home page and introduction page improvements. My design workflow requires deep remote collaboration with the global team and has taught me how to best work with teams located around the world.

User flows and wireframes

Exploring paths forward

With any new feature for AI Builder there is a lot of exploration prior to final designs being developed. For larger efforts, I often led brainstorm sessions with our broader team to gather innovative ideas outside of our design group. Userflows and wireframes are a crucial step to help the team arrive at the most successful design direction.

AI Builder brainstorm Conducting a brainstorm session for a new AI template. AI Builder user flow AI Builder user flow Userflow details from early drafts of the Prediction template.
AI Builder wireframe
AI Builder wireframe
AI Builder wireframe Early wireframe explorations for discovering and creating an AI model.

Research

What do users want and need?

AI Builder is fortunate to have a dedicated researcher to conduct user studies, create benchmark reports, and gather feedback from customers. Design research is always a key element which shapes and informs the direction for our design work.

AI Builder persona map AI Builder persona tools Persona maps paint a clear picture of the journey our users go through.
AI Builder research insight
AI Builder research insight
AI Builder research insight
AI Builder research insight
Research insights from Prediction model user studies.

The launch

AI Builder is out in the wild

AI Builder was released to the public in June 2019 and started with just a handful of templates. The public’s reaction and general excitement for AI Builder was very encouraging and motivated us to work even harder to improve AI Builder. As new challenges arise, our team continues to leverage research and our iterative approach to ensure that we create the experiences our customers want and love. AI Builder has brought measured value to the Power Platform business and continually extends the possibilities of Power Apps and Power Automate.

AI Builder launch
AI Builder announced at the Microsoft Business Applications Summit - June 2019.
AI Builder Build page
AI Builder home page with all AI templates and prebuilt models to explore.
AI Builder Get Started
AI Builder Get Started
AI template modal window with instructions for creating and using the model.
AI Builder Wizard
The AI Builder Prediction model multi-step 'wizard' to help users create predictions based on historical data.
AI Builder Wizard AI Builder model details page
After a model is trained, the details page allows the user to review results and then use it in an app or a flow.
AI Builder Studio
Trained and published models are available to add to apps created in the Power Apps studio.
AI Builder Business Card App
AI Builder Business Card App
AI Builder Business Card App
Business Card Reader app example with built in AI Builder model extracting contact information from a business card photo.

Design process example

Form Processing design sprint

To further illustrate my design process from conception to final design, I have outlined an example feature our team executed recently. Although design is rarely a linear process, these steps help describe how I approach creating a design that our customers want.

Form Processing is an important AI Builder template that can be trained to automatically extract, organize, and save data from routine documents such as invoices and tax forms. From research and user feedback it became clear that Form Processing needed to allow for more user control. In addition to the auto detected fields in a form, users wanted control to select and edit the areas for the model themselves. The mixture of auto and user generated areas required a lot of exploration and iteration. In order to arrive at a solution quickly we ran a week long design sprint that ended with a successful research study and clear path forward.

AI Builder sketch Our team held a brainstorming session to generate as many new ideas and directions as possible. After voting, two possible directions were selected to test with users. AI Builder sketch My initial sketches focused on how to educate the user on the new feature and to lead them through the required selection steps. AI Builder whiteboard I led a group white boarding session to prove out the chosen direction and to fill in the gaps of the user flow. AI Builder user flow Once we settled on the flow, our team brought the concept into Figma and collaboratively worked on pulling all the required screens together for a complete flow. AI Builder user flow To help the design come alive we built a working prototype with Figma to test with customers. AI Builder user study Within a day, our prototype was ready to test and our research team presented it to a handful of users to gather feedback and reactions. Overall, user reception was very positive! So, with a few adjustments we moved forward with our direction and were ready to create the final designs. AI Builder user study The second version of Form Processing was released in the spring of 2020 and was a welcomed update from customers eager for the feature.
AI Builder Forms wizard
AI Builder Forms wizard
AI Builder Forms wizard
AI Builder Forms wizard
AI Builder wizard AI Builder model details page