Building a conversational interface trusted by AI sceptics
IN Partnership with
Everyone is talking about AI. But in financial services, the biggest challenge isn’t whether the technology can work − it’s whether people will trust it
More
WHEN IT comes to insurance, mortgages and financial advice, trust isn’t a nice-to-have. It’s the foundation of every interaction. Customers trust advisers to guide them through important decisions, and advisers need confidence that any technology they introduce will strengthen that relationship rather than weaken it.
So, at Source Insurance, when we began developing a new AI-powered approach to home and landlord insurance referrals, we started from a position that may seem counterintuitive: scepticism. Not because we doubted what AI could do, but because we understood what it needed to overcome.
When you go out in the rain, you don’t wear flip flops and hope for the best. You dress for the conditions. Building AI in
Source Insurance is an award-winning home and landlord insurance intermediary, partnering with advisers across the UK financial services market. With over 30 years’ experience, Source delivers high-quality insurance solutions through a trusted panel of leading insurers. Their intelligent and flexible approach allows advisers to “choose their easy”, with both fully advised and referral routes available for arranging home insurance. With a 5★ Defaqto accredited panel and 5★ Trustpilot rating, Source is committed to exceptional service. Working closely with advisers and customers, Source continually evolves its products and services through direct feedback to meet market needs.
Find out more
Sceptics shape better
AI design
One of the most valuable lessons we’ve learned is that the best feedback rarely comes from people who evangelically believe in AI. It comes from experts who hold a healthy wariness and can certainly scrutinise it.
“Referral solutions never quite solved the hardest part of a GI referral … keeping customers engaged once the adviser has handed over. You can’t meet the customer’s need if the customer has logged off”
Chris Lynch,
Source Insurance
financial services requires the same mindset. You don’t build for the perfect scenario; you build for the moments where confidence drops, conversations become complex and human judgement matters most. So, with our scepticism wellies firmly on, we began.
Trust starts with solving the right problem The challenge we wanted to solve wasn’t new. For years, general insurance providers have tried to make referrals simpler through “hands-free” referral models, reducing administration for advisers and creating a smoother customer journey. On paper, the approach made perfect sense.
“The problem was that referral solutions never quite solved the hardest part of a GI referral,” says Chris Lynch, IT director at Source Insurance. “It’s not making the referral. It’s keeping customers engaged once the adviser has handed over. You can’t meet the customer’s need if the customer has logged off.”
Customers found themselves pushed into unfamiliar portals, online forms or email journeys, and that is exactly where engagement falls.
We knew the answer wasn’t another portal. It was a conversation. Not a conversation with AI for the sake of it, but one that enables customers to complete a home insurance quote through a simple back-and-forth exchange. Instead of asking customers to translate their circumstances into rigid form fields, the conversation guides them chronologically through the process one question at a time, rephrases anything that’s unclear and adapts based on their responses.
This approach reflects how customers increasingly expect to interact with digital services. Messaging has become second
nature, while conversational AI interfaces such as ChatGPT, Claude and Gemini have made interacting with AI increasingly familiar.
The opportunity wasn’t simply to automate the quoting process. It was to create an experience that feels more natural and intuitive, helping customers complete their quote with confidence while remaining in control throughout.
However, familiarity alone doesn’t create trust. Trust must be designed into every interaction, which meant asking ourselves a different question: not “How much can AI do?” but “Where should AI stop?”
Building trust in public We knew that we couldn’t answer that question in a meeting room. If trust is earned through experience, then the people using it need to shape that experience. So rather than building quietly behind closed doors, we’ve chosen to develop this conversational interface alongside advisers and customers. We’ve deliberately exposed our assumptions early, invited challenge and used that feedback to shape the product.
One of our closest partners throughout this process has been Connect Mortgages.
“Connect Mortgages are an integral partner for us as we continue to beta test and refine,” explains Lynch. “Together, we’ve deliberately pushed the system to its limits. We’ve tried to find the moments where it falls short, where the experience feels unnatural and where human involvement is needed.
“Those challenges have helped us build stronger boundaries and helped us understand where to
“We’ve been through multiple test cycles with Source. We’ve delivered our feedback, talked through solutions and they’ve gone away, worked on improvements and delivered a new version for us to test again”
Liz Syms,
Connect Mortgages Limited
add hard-coded guardrails to continually improve the experience.”
Key questions we have asked ourselves through the process are:
Where should the AI respond?
Where should it ask for clarification?
When should it explain that it isn’t certain?
When should it stop altogether and introduce a human adviser?
And these aren’t just technical decisions. They’re trust decisions.
Published August 24, 2026
Share
Trust will define AI’s future
The future of AI in financial services will not be built by convincing everyone that the technology is perfect. It will be built by creating solutions, where we understand why people are sceptical, respect those concerns and earn trust one interaction at a time.
Our sceptics have become our best designers
One of the most valuable lessons we’ve learned is that the best feedback rarely comes from people who evangelically believe in AI. It comes from experts who hold a healthy wariness and can certainly scrutinise it. Every round of testing challenges another assumption. Every conversation helps us understand where the experience feels natural and where it feels forced. That iterative process has become central to how we’re building the platform.
“We’ve been through multiple test cycles with Source,” says Liz Syms, CEO of Connect Mortgages. “We’ve delivered our feedback, talked through solutions and they’ve gone away, worked on improvements and delivered a new version for us to test again. Advisers having direct input into a solution they will be using is essential for adoption. They are the ultimate test group.”
Rather than seeing scepticism as resistance, we’ve used it as one of our greatest design tools.
What we’re learning We’re still refining the experience with Connect Mortgages, but what is clear is confidence is earned through repeated challenge. We’re still challenging ourselves and they’re still
challenging us and that’s exactly what we want. Perhaps the most encouraging sign hasn’t been the technical performance of the system. It has been watching people who were openly sceptical of AI begin to trust the experience.
Because the future of AI in financial services will not be built by convincing everyone that the technology is perfect. It will be built by creating solutions, where we understand why people are sceptical, respect those concerns and earn trust one interaction at a time.
We’re proud to be AI sceptics.