A shift from too many options to guiding decisions
What began as an exercise to integrate new UI components into the current live-build revealed that simply adding new units compounds existing issues of high visual density and cognitive load. The current structure prioritises content exposure and browsing, forcing users into an exhausting exploratory mindset rather than guiding them toward confident decisions. Because the fundamental limitation lies in the architecture rather than the components themselves, the project’s focus shifted from component design to rethinking the homescreen’s underlying decision model to better drive conversion and user satisfaction.
Position
I started by rebuilding the current app in Figma to see how new UI components, like the "Spotlight" unit, would fit into the live layout
The main goal was to see if adding these new features actually made the experience better or just more cluttered
Problems
Too much noise: Adding new features to the old structure just made the screen feel messy and harder to use
Information overload: It’s forcing users to scan and hunt through a sea of content, which makes it exhausting to figure out what to do next
Wrong setup: The current layout is built for endless browsing, which gets in the way of people trying to make quick, easy decisions
Possibilities
Fix the foundation: Instead of just tacking on new components, we have the chance to rethink how the whole home screen is built
Guide the user: We can shift the focus from "just looking" to helping people actually get things done
Better results: By making the path to a decision clearer, we'll make the app feel more intuitive and boost conversion at the same time
Re-positioning statement…
From Content Exposure
To Decision Guidance
What the current design does…
The current homescreen follows an experience flow of: Promote → Browse → Browse more.
It prioritises surfacing promotions, categories, and partner content, giving users many ways to explore the platform.
Where it works:
Supports discovery and variety
Maximises partner exposure and visibility
Allows users to browse freely across multiple missions
Where it falls short:
Lacks a clear path to decision
Mixes promotion, browsing, and decision-making in one flow
Requires users to scan, interpret, and compare before acting
Conclusion:
It works well as a discovery and exposure surface, but less well as a tool for making quick, confident / fast decisions.
Customer needs and expectations…
From customer needs to insights
Concept 1. Conversational AI, Financial Assistant (day 1 to 28+).
This concept reimagines how a conversational AI-powered Financial Assistant could transform customers' attitudes and behaviours toward their money and financial future. Accessible through voice commands or a combination of voice and minimal touch-screen interactions, this assistant evolves with use.
Initially, the AI provides introductory guidance to establish a foundation. Over time, as it learns the user's preferences and patterns, it transitions to offering concierge-level service, effectively managing existing financial products while educating customers and helping them confidently plan their financial future.
NOTE: To experience prototypes properly, please enable your “speakers” and follow the instructions below.
Click image to Play prototype / Click to Pause
Click image to Play prototype / Click to Pause
Concept 2. AI Concierge + AI Assistant + AI Scanner.
In this third iteration, AI can also exist in vertical and horizontal business channels, servicing specific products e.g. Mortgages or Insurance to help get users informed faster and increase conversion rates.
We’ll need an AI multi-agent OS sitemap to make this work
(customer-facing UI at bottom of diagram)
To maintain an efficient service, comparable to that of a human agent, the LLM must be lightweight, fast and capable of making informed decisions based on fast learning and a support network reinforcing the knowledge database via a human trainer.
In the diagram below, a host of AI agents provide background support and learning capabilities to a single customer-facing touchpoint .
Two interesting observations arose from this project. Firstly, 99% of people that I consulted, and who saw the final presentation, were ‘not’ aware of conversational AI’s capabilities. Many thought ChatBot’s were the pinnacle of such systems. Secondly, the overwhelming feeling, at the end of the presentation, was that improving customer knowledge would increase a customer’s confidence and engagement with their finances. Coupled with the nurturing assistance of a Financial AI, it was felt that this concept could be the perfect conduit needed to tie various parts of the business together, provide a much better customer experience and, increase revenue.
This concept is now being further explored by Lloyds’ dedicated AI team.