Watson Discovery Service
Software Product Design
Lead Product Designer
Watson Discovery Service is a cloud based software that enables users to rapidly ingest unstructured data, and convert that data to a structured JSON output while enriching it with cognitive APIs. These cognitive APIs enable Discovery Service to uncover deep connections throughout data by way of advanced AI techniques. The Watson Discovery Service tool was released to the general public in late 2016.
The following case study is personal and does not necessarily represent IBM’s positions, strategies or opinions. I have omitted and obfuscated confidential information.
I was one of three UX designers working to launch Watson Discovery Service in 2016. After launch, I became a Lead Product/Feature Designer focusing on evolving the service through the definition and execution of future features and enhancements, while simultaneously addressing customer pain‐points through regular usability testing and continuous delivery cycles.
UX Designer/ Lead Product Designer
2016 - Present
Developer Tool, Enterprise Software, Artificial Intelligence
Our challenge with Watson Discovery Service was to create a complex developer tool which utilizes a human-centered design without limiting the power of the technology.
Intelligent systems, by their nature, can be very complex. Oftentimes tools meant to help users create intelligent systems are themselves too complicated for the user to work with. Vice versa, sometimes these tools become oversimplified in an attempt to aid the user and inadvertently limit the power of the technology.
The goal of Watson Discovery Service was to create a tool to be utilized by developers of all backgrounds to help them easily leverage the technology needed to create intelligent systems and applications of their own.
Squads & Interdisciplinary collaboration
Interdisciplinary collaboration and the utilization of squads was an integral part of the design process for Watson Discovery Service. Given the size and complexity of the product, several separate squads had to work together to bring multiple micro-services together into one cohesive tool. These squads were comprised of Offering Managers, Developers, QA Engineers, Researchers, and Documentation Writers.
To enable this collaboration, myself and the other two designers working on the product hosted a series of workshops to bring members of the various squads together. These collaborative workshops accomplished several goals:
1. Empower all members of the team to have a voice and share ideas
2. Educate all team members on the nuances and inner- workings of the various technologies being integrated together
3. Collaboratively break down problems the service faced and ideate on possible solutions
4. Keep the team empathizing with and focused on the needs of the user
Rapid prototyping was another essential part of the Watson Discovery Service design process. One of the primary outputs of the collaborative workshops were ideas and sketches that were later turned into prototypes. These concepts would then be wireframed in Sketch, and then transformed into a clickable prototypes in Azure. From there the prototypes would be tested.
Usability testing for Watson Discovery service operated on a 3-week cycle. Every 3 weeks our team Researcher would compile our latest workflow concepts test them with real users in an interview style remote testing session. All members of the squads were encouraged to observe these testing sessions and participate in a debrief meeting after the testing was complete.
For simple interaction-based concepts, we tested our prototypes with various online testing tools on an ad-hoc basis.
All of these practices comprised our Loop. The Loop is a model in IBM Design Thinking which represents a continuous cycle of observing, reflecting, and making. As our squads moved through sprint cycles, we would incorporate collaborative workshops, rapid prototyping, and usability testing to propel us through the loop, until we were ready for our GA release in December of 2016.
Watson Discovery Service tooling was our solution. The tooling provides an interface which allows users to upload their data, enrich it with cognitive APIs, normalize it into a JSON format, and then query it to product API calls to be leveraged in the user’s end applications.
Watson Discovery Service was released to the general public on December 15th, 2016. The service is used to power applications for customers all over the world from ESPN to HR Block, and Salesforce.
One of the first features designed and implemented into Watson Discovery Service post-GA release was Insight Cards. After launch, users began providing feedback that indicated that they were unsure of what the cognitive APIs applied to their data enabled them to do.
After a series of brainstorming sessions, prototypes, and usability testing, the team aligned on the concept of insight cards. Insight Cards are a set of UI elements which display key insights from the user’s newly enriched data. Insight cards begin to query the user’s data as it is being ingested into Discovery Service. They then display the top entities, concepts, keywords, hierarchy, and sentiment present in the data. Several rounds of usability testing and interviews were performed to refine the concept before finally releasing the enhancement in the product.
Insight Cards to date have been one of the most lauded features on Watson Discovery service. Users are now able to see, at a glance, top-level insights from their data in real time and gain an understanding of the power of Watson’s cognitive APIs.
Following Insight Cards, we integrated another feature called the Schema Explorer into Watson Discovery Service. While performing usability testing and interviews for the Insight Cards feature, it was revealed that users were also unclear as to what the final format of their data looked like after ingesting it into Watson Discovery Service. This hindered users from being able to effectively query their data.
The initial concept for Schema Explorer was the product of a Watson Discovery Service 18 month vision workshop. The workshop took place over several days and involved all members of the product’s design team, as well as senior design directors from the Watson team. Following several days of collaborative ideation and empathy mapping, the first schema explorer concept had been created.
Several sprints of ideation, prototyping, and testings later, the Schema Explorer was launched. With these new features users are able to view the newly structured fields in their data and how the Watson enrichments have been applied to them. They are also able to run sample queries directly from the schema explorer to kick off their querying experience.
Watson Knowledge Graph was a technological enhancement that allows users to make connections across documents and generate new knowledge. The Knowledge Graph technology was created by a research team located in another area of the country. The process of bringing the technology into Watson Discovery Service involved intimate coordination and collaboration between remote teams.
I organized a series of remote educational and collaborative workshops with each respective team’s Offering Managers, Developers, and Researchers. Over the course of several days we aligned on a user story, and mapped the user’s journey from within Watson Discovery Service. The teams were able to agree on an approach and formulate a plan for integration into Watson Discovery Service. Today the feature is available to Advanced and Premium plan users.