When Adobe Customer Journey Analytics users want fresh insights from their data, there’s a new assistant that can help—and it’s AI powered.
The Adobe Research team recently collaborated with their counterparts on the Customer Journey Analytics product team to develop a built-in Assistant. The tool allows customers to describe what they want to analyze in natural language—and then the Assistant instantly creates a visualization. Users can dig deeper by modifying their initial prompts, such as adding different attributes or time ranges, to get a new view.
“The new Assistant lets people overcome the difficult and time-consuming process of analyzing data and creating visualizations using complicated user interfaces. You used to have to browse through thousands of attributes to select the ones that interested you,” says Ryan Rossi, a research scientist at Adobe Research and one of the key contributors to the new Assistant. “Now, users can simply say what they want in their own words and see it visualized.”
The tool was designed to improve productivity for Customer Journey Analytics users while helping them find new insights. And since it makes exploring data quicker and easier, customers will have more time to tap into data they wouldn’t normally analyze, giving them even more information for making business decisions.
The research behind Adobe’s custom generative AI
At the heart of the new Assistant is Adobe’s own custom text-to-visualization generative AI solution.
The model first decodes natural language queries, such as “compare revenue month over month by country,” into more structured outputs the system can use, including the user’s intent, corresponding metrics, a time range, and other information that supports Customer Journey Analytics’ capabilities. Then, the model turns this information into a data query and generates a visualization that answers the precise question the customer asked.
The process behind the scenes is complex, but the end result is a simpler user experience.
“So far, we’ve heard very positive feedback from users who’ve tried the Assistant,” says Rossi. “One even called it ‘indistinguishable from magic.’ Others told us it’s been super-easy to use, and that it allowed them to quickly see trends and explore important metrics. Another user told us that they’ve found new insights they would never have even explored with the old, manual method.”
The Assistant was developed with contributions from members of the Adobe Research team, including Ryan Rossi, Sungchul Kim, Tung Mai, Tong Yu, and Ritwik Sinha.
Wondering what else is happening inside Adobe Research? Check out our latest news here.
