The Document Intelligence application was being developed by EY’s Artificial Intelligence Lab in their Palo Alto office. I was brought on to the project as an additional UX resource to help with redesigning the application. At the point I was brought into the project the entire application had basically been built by the development team and was being tested. Needless to say the user experience of the application was in need of some help.
The team was located in various locations, myself in Alpharetta GA, the development team in Palo Alto and Mexico, the product owner in Palo Alto and the other UX Resource in Argentina. After about 1 month the Argentina UX resource was moved to another project and a UX Designer with visual design experience was added to the project in Palo alto. My strength coming in the form of the research and strategy areas as well as UX architecture.
Because the application had already been built the UX process was one of just iterating through improvements to the UI and trying to make improvements as best we could. Primarily myself, the other UX resource and the Product Owner would meet twice a week to ideate on the various sceens of the application
The prodcut owner did recruit some users to test the application. The test consisted of having them all use the same lease contract, uploading it to the application. Then the users created a lease profile, let the AI run the profile against the document and then try to extract the information they would normally extract. Our big part of the test was the feedback session we'd record with the users where they would walk us through the process they used to complete the test.
The project was basically a drink from the firehose introduction and trying to digest what the application did and the different areas of the application. The main purpose of the application was to take legal contracts, create profiles for the types of contracts and then allow the AI back end to use the profile to pull out the parts of the contract that would normally be pulled out manually by and EY employee. While years of experience helped EY employees with knowing the structure and content of a contract there was a substantial number of new or less experienced employees that spent a lot of time going through contracts and the application was being created to help reduce their time finding the specific information they were looking for.
At this point in the project the leases being focused on were real estate leases. So an example of the kinds of information the employee and AI would be looking for would be:
The discovery process also allowed me to learn that not just one document made up the process. More than likely throughout the contract creation cycle the master contract would have amendments made to it which meant other contract files. The application would be able to create contract groups that would be comprised of the master contract and all of its amendments.
To help with how the application was structured as well as creating a mental model for myself I created a concept map, a sitemap and the workflow for adding file and creating file associations (i.e. master/amendments) for the application.
With and understanding of the application I worked with my UX counterpart and the product owner to brainstorm some ideas out for each of the current screens. From our meetings I would work up some examples for possible solutions, then we’d meet evaluate the wires, suggest improvements and then make updates. We’d meet twice a week on Tuesdays and Thursdays to discuss the status of the wireframes. Below are some of the examples of the wireframes, these images show the overall main screens but not all of the interactions that each screen could have.
From the sketch files we created clickable prototypes in InVision. The usability tests focused on the Users interface (adding and removing users from the application) and the Master and Amendment associations. For the usability tests I created a test script with tasks asking the users to complete tasks like:
YES, there are a lot of images below because I wanted to show just how many images I had to create for the clickable prototype InVision (and keep track of for creating hotspots and connecting images in the right order). This is frankly why I prefer Axure when it comes to creating prototypes. It's easier for me to make updates and change the prototype. I love Sketch too for creating wires etc. but there are just things I find easier in Axure.
My time on the project was a 4 month time frame until I returned to the experience design group to move on to other work. Overall I feel I contributed well to making improvements to the UI. The design and development of the application is still ongoing as of today but my contributions were echoed by the product owner after I finished up working with the AI Labs group…
“Mike has been really instrumental in hitting our milestones and tackling challenging UX and UI issues we face uniquely in AI tools. He always brings new ideas to the table and keeps his time commitments with good quality work. He is very collaborative in nature and has taken lead in redoing UI concepts for us that we will doing usability testing with. He has been key in instrumenting how we think about UX and do some user testing before committing to big changes. It has been great working with him.” - Nischal , Document Intelligence Product Owner