Smart Insights in Transactions
Designing an AI-powered transaction search that keeps users in control.
Client:
Candescent
Category:
Product Design, AI
My Role:
Product Designer
Adding AI powered insights to transaction search without disrupting core tasks
Introducing AI-powered insights into transactions for Candescent's new mobile banking white label offering.
When users open their transaction history, they usually have a specific task in mind. They want to find a purchase, check what they spent with a vendor, track down a refund, or compare subscription costs. Exploration isn't what brings the user in.
I was brought in to design an AI insights experience for a new banking app that was still being built. There were limited product artifacts and no established patterns to work from, so I worked closely with the team building the new app while figuring out how AI could fit into the experience without stepping on the core functionality.
The business ask was to create an optional path into conversational AI. My challenge was figuring out how to do that without turning a straightforward transaction-search task into an AI experience.
The Approach
AI should add context when it’s useful, but the core transaction workflow stays fast, familiar, and task-focused.
I started by looking at the different things people might actually want to find in their transactions. User want to see spending within a timeframe, purchases from a specific vendor, subscriptions, refunds, payment history, etc.
I explored several approaches for surfacing AI, including more prominent insight cards with action-oriented CTAs, before landing on a simpler solution. The search stays the primary experience, and AI becomes an optional layer on top of it
Users can search with a keyword or full question.
The transaction table updates directly based on their search, similar to a filter.
A small Smart Insight appears above the results with relevant context.
Users can dismiss it, ignore it, or select it to learn more.
Selecting the insight opens a conversational layer with contextual follow-up suggestions, such as comparing spending or taking action on a subscription.
The result is a simple progression: search → insight → deeper exploration, with the user deciding how far they want to go.
Insights are used to enhance core functionality. Users have the option to dig deeper, learn from a glance, or skip it.
Related Projects
