In October 2023, our work at UiPath was recognized when Clipboard AI was celebrated as one of Time Magazine's Inventions of the year. I am thrilled to share this milestone with our dedicated team at UiPath and all those who contributed to this remarkable product.
Image from Time.com
We started with an initial idea, semantic automation. Automating from user intent rather than using programming at all. A couple of hackathons later and we had a proof of concept that was really compelling.
Automation at UiPath is often about moving data fluidly between many different systems. Systems that don’t easily connect, don’t have available APIs or just aren’t important enough to attract the attention of IT departments. Despite the work we’ve already done to automate so much of this, we knew that people everywhere still spent a disturbing amount of their lives copying data from one system and pasting it into another.
It seems to be the curse of design: when a product is completed, it seems like it was an obvious thing, the decisions that were made along the way coalesce into a solution that “just makes sense”. I can assure you we didn’t start that way. Our first attempts were good, but they weren’t enough to incentivize users to do their work in a new way. We couldn’t just be better, we had to change their way of thinking. Copy pasting was ingrained, and we were aiming for a 10x improvement in overall productivity in these scenarios, a number we felt would trigger viral behaviors.
The answers didn't come easy; the pixels and stars didn't align perfectly, and we were far from knowing what functionality needed to be in the product from the start.
Clipboard AI is the result of relentless dedication to understanding our users' needs. We embarked on an ambitious project, conducting 24 unique research studies (so far) to understand our users deeply. Diary studies, RITE usability studies, naming studies (can you believe it was originally called “Semantic Copy Paste”?), and generative observational studies were used to guide us to a product our customers would use.
Image from Time.com
While executing these studies, we found that nearly without exception data was moved out of its originating system, into a spreadsheet where it was reviewed and then copied to the target system. Different jobs had different tolerances for error. In HR, errors were more accepted, though still frowned upon. In finance and accounting, an error, even if it occurred less often than when the data was manually entered, was considered an immediate blocker.
In these studies, we discovered that technical limitations that we had accepted initially would need to be overcome for us to be successful. In the initial implementations, the technology couldn’t read the page beneath the fold. We assumed that our users would accept that they were only pasting what they could see. Through our diary studies, we were quickly corrected, learning that our customers expected more.
It didn’t just take a lot of research, the design team explored hundreds of variations and directions to find our way.
Every innovation faces hurdles. For Clipboard AI, it was about tackling these technical “limitations” head-on. We knew that a clipboard tool that's truly powerful should be seamless, and this meant solving for the technical issues that engineering had determined was “impossible”. We proved through our research that these solutions were necessary, and engineering rose to the challenge and solved dozens of these impossible problems.
So how do you go from a "good idea" to "Invention of the year"?
Start with Humility
At UiPath, humility is not just a buzzword; it's a practice. We learned that even though we had good ideas, we needed to listen to our customers to understand how those ideas fit into their daily work.
By listening to our customers we learned that validating the work done by the AI was important. So we created an intermediate step where they could look and see what was about to be pasted. We didn’t listen to their solutions, we paid special attention to their problems, allowing us to build something better than the world they lived in.
Pick the right Research Approach(es)
To respond swiftly and efficiently, we adopted RITE (Rapid Iterative Testing and Evaluation) as our model for usability testing. This allowed us to discover and address issues at an extraordinary speed.
The RITE usability method means we don’t have to wait for the report to start solving our customer issues. We put our product in front of customers and fix what blocks them that day. Then we test it again. By iterating quickly we can fix significantly more issues, and our product becomes much easier to use in a shorter time frame, increasing overall team agility.
Extensive Outreach
In this process, the design team has engaged with more than 250 individuals during the tests alone. These diverse perspectives enriched our understanding of user needs and preferences, helping us fine-tune Clipboard AI to solve their problems.
Problem-Solving
We've all heard the phrase, "That's impossible," when engineers were confronted with specific goals that were extremely hard. However, we learned that what may seem impossible is often just a difficult challenge. We stuck to our commitment to fixing the impossible, ensuring that the challenges were on our side, not our customers'. We pushed hard to have these solutions implemented and Engineering rose to the challenge. Never accept “It’s impossible” for an answer.
The Importance of Adaptation
In the world of design and innovation, research should not be a one-time effort; it should be an ongoing journey. When the data takes you in a different direction, follow it. Your customers are your guiding light, and they know what they're doing. Stay humble, stay curious, and stay adaptable. The path from a "good idea" to "Invention of the year" is paved with discovery, delight, and dedication.