Google AI (with Orby) has been launched in May, 2023.
.png)
Organizations are inundated with an overwhelming volume of documents that need to be processed, validated, and organized. Orby AI, powered by Google Document AI, helps customers automate document-centric repetitive tasks like contract processing, email processing, and invoice validation.
Contract project
As a Design Associate, I solved the key usability issues and redesign the experience of the review and validation process.
I worked in a cross-functional workshop, collaborated with stakeholders, developers, and project managers to ensure successful delivery.
From 45 minutes to 9 minutes 🚀
End-user testing with our key users showed the time to collect information and read the contracts dropped from 45 min to 9 min. It has dramatically reduced the data collection and aggregation burden on finance teams. Our product help increase the speed of processing contract by 80%.
.png)
.png)
A User-friendly
Tutorial
it decreases training time and the learning curve
Users find it boring to have to train over 100 contracts before being able to use the AI system.
A Goal-oriented
Suggestion List
Access to Important Information
Users struggle to locate the task file within hundreds of tasks in systems because the files names are almost same
An Accessible
Review Page
It gives end users full control and enhance their confidence
Users do not trust the accuracy of AI when they review AI’s extracted results
Financial professionals are known for their meticulous work ethic, where even a single word cannot be extracted inaccurately
Everyday, they open up google drive to view new contracts waiting to be dealt with. Then they review contracts, invoice, and other financial documents. They extract new terms and put them manually in a master spreed sheet.
.png)
How might we help finance operation specialists review and process contracts more efficiently with AI automation?
The financial specialists work starts with
training Orby algorithm,
locating the documents Orby manages,
and finally extracting terms in the contracts
.png)
Based on feedback from client engineers, product managers, and usability testing. We conducted iterations, including user flow, wireframes, and visual presentation
Implementing a Training mode editor
.png)
"I don't realize Orby is learning behind the scene when I extract terms when I am inputting them into Sheets."
"I get familiar with the interfaces while I can quickly train Orby system by selecting keywords in the box. After which they can be automatically fill in the entities."
.png)
.png)
Identify different files with complex information
.png)
“I need more info because I need to know which file should be dealt with first"
"I identify different files with complex information including name, file ower, file type. last updatad time.
.png)
.png)
"It does not match my working habit like add missing terms and reject terms"
"All the information presents equally in the list"
"It is so easy now to check the extraction results one by one by tapping "enter" on the keyboard"
"I can tell which words may be misrecognized"
.png)
.png)
.png)
Rooted in user needs, gather design ideas from real use cases.
The team needs to engage in open communication with the engineers to address the concerns. Explore useful solutions. Be a resourceful designer and find different ways to solve problems.
Promote an iterative design process where improvements are continuously made based on user testing and feedback. Remember that a design is never really 'finished'.
