A conversational AI that guides customers to the right products
Creating a trusted AI companion that helps customers compare products, understand trade-offs and take confident next steps across three MONY Group brands.
Role
Product Designer
INDUSTRY
Financial Services & AI
Duration
18 months
Tools
Figma, FigJam, Make, GPT, Dscout, Dovetail
PROBLEM STATEMENT
MoneySuperMarket early AI assistant behaved like a standalone chat tool: it had limited memory, little awareness of customer context and no consistent role across brands. The goal was to create a trusted companion that could guide customers to relevant products, explain options clearly and act only when it had the right level of confidence.
THE TEAM & MY CONTRIBUTIONS
As the sole product designer, I partnered with Heads of Product and a team of 5 AI engineers to define the product strategy, interaction patterns and behaviour rules that made the Companion reliable, useful and scalable.
Evidence & Reasoning
Understanding how AI shapes financial decisions
Diary studies and user interviews explored how people use AI for shopping, financial planning and everyday decisions. The research showed that customers already turn to AI to reduce effort, but need clear reasoning and relevant context before they trust a recommendation or act on it.
Discoveries & Opportunities
1. Be more than a conversational experience
The Companion should help customers make decisions, not simply return results. It needs to explain why a product fits, surface meaningful trade-offs and make the next step feel clear.
2. Personable and context aware
Customers expect the experience to remember information they have already shared and adapt its guidance to their goals, profile and stage in the journey.
3. AI is already part of their research
Many customers already use tools such as ChatGPT and Gemini to narrow options. The opportunity was to bring that support into a trusted, brand-owned experience with product data and service expertise.
4. Extend into financial wellbeing
What-if scenarios can make the long-term value of switching more tangible, helping customers understand potential savings and choose with greater confidence.
STRATEGY & APPROACH
System & Behaviour Design
The strategy moved the experience beyond chat by defining three AI roles. Each role specifies the customer goal, trigger condition, information required, user action, AI response and guardrails, so the system stays useful without taking more control than the moment warrants.
Assistant (Low Autonomy)
The Assistant removes small points of friction without interrupting the journey. It offers inline suggestions, smart defaults and contextual hints while leaving the decision and action with the customer.
Analyst (Medium Autonomy)
The Analyst supports decisions that need explanation. It asks focused follow-up questions, compares trade-offs and turns complex information into price insights, summaries and ranked options.
Operator (High Autonomy)
The Operator completes well-defined tasks end to end when the customer has delegated the work. It can optimise prices, re-quote or make automated updates, while clearly showing what changed and why.
OUTCOMES & DELIVERABLES
Product & Agents
The framework was applied across real customer experiences—from a third-party AI ecosystem to MoneySuperMarket’s own web and app—so each product used the same principles for trust, autonomy and clear decision support.
ChatGPT Connector App
First financial ChatGPT app launched in the UK
The first financial ChatGPT app launched in the UK. It brought MoneySuperMarket’s product expertise into the ChatGPT ecosystem, helping customers explore options through a consistent brand voice and clear service value.
Price Optimiser
An AI-powered car-insurance feature that uses MOT information and customer details to test relevant quote variations. It identifies meaningful savings, explains the parameters it changed and presents only the options worth considering—demonstrating the Operator role in a high-value task.
Other agents
Broadband Agent: helps customers identify broadband deals that fit their household needs, usage and budget.
Credit Score Agent: connects customers with relevant products and practical guidance to understand and improve their credit score.
IMPACT
Key Learnings
AI UX starts with customer intent, mental models and clear guardrails—not the interface alone.
Trust grows when the system explains its reasoning, respects customer control and knows when to ask before acting.
Impact
Price Optimiser engaged 10% of eligible customers; 60% of those customers found a cheaper car-insurance offer. This demonstrated the value of applying trusted, high-autonomy AI to a concrete customer goal.
