AI, Customer Service Transformation, Voice
ID&V, Reimagined: How 1st Central Modernized Customer Verification with Gen AI
1st Central is one of the UK’s fastest-growing insurance providers with around 1.5 million customers. Focal to 1st Central’s business strategy is using technology, including data analytics, AI, and digital capabilities, to deliver underwriting excellence and both customer experience and operational efficiency.
This technology-driven approach has enabled the company to deliver effortless customer experiences and great value resulting in an ‘Excellent’ Trustpilot rating.
An Interview with 1st Central on Integrating Generative AI into their Retail Operations Contact Centre
I recently had a virtual conversation with Andrew Smith, Customer Director at 1st Central, discussing their integration of Generative AI (Gen AI) and Large Language Models (LLMs) into its contact centre operations. We explored the approach 1st Central took to adopting these AI technologies, the use cases it identified that would have tangible outcomes, and the successes achieved so far, not only in terms of customer experience but also for frontline colleagues.
Through an introduction from our partner Opus Technology, ServisBOT has been privileged to collaborate with 1st Central since 2023, supporting it with the selection and deployment of appropriate use cases for LLM and Gen AI technologies in its retail operations.
The story is best told in Andrew’s words as he describes 1st Central’s first use case as it shifted from manual to automated identity and verification (ID&V) using generative AI:
Kristen: Andrew, thank you for taking the time to chat with me today. Let’s start with a brief introduction to your role and responsibilities at 1st Central with some context on the scale of your contact centre operations.
Andrew: I’m the Customer Director at 1st Central. I manage the technology which powers our contact centers – telephony, chat, email and social – so customers can contact us using their preferred communication channel.
Our part of the front line is responsible for anything that has to do with sales, renewals, making changes to a policy, missed payments, and other general queries not related to a customer’s claim. Our retail operations contact center is split geographically across three locations—operating in the UK and South Africa.
THE INITIAL USE CASE FOR GENERATIVE AI: AUTOMATED ID&V
Kristen: Could you share why you selected your initial use case, what issue you were trying to solve, and how you went about the design and implementation?
Andrew: We started with identity and verification (ID&V) as our first voice AI use case.Our goal was to provide a more seamless ID&V experience for those customers that want to engage through the voice channel. For a traditional ID&V process, there is a limit to how streamlined you can make it. We must capture a certain amount of data from customers so that we can verify their identity.
We had tried to streamline this process using better scripting and undertaking extra training to make sure that the frontline staff were handling ID&V as efficiently as possible. But the missing link was always the technology.
Previously, when a customer called us, a frontline colleague would ask the same ID&V questions at the start of each conversation. They would then register the responses and check them against the information in our system.
This is often seen as the tedious part of any interaction with a call center. And if it’s boring for customers you can bet it is for the agent too!
So we thought, imagine if we could eliminate this mundane part while simultaneously maintaining—or even enhancing—the customer experience. This would allow customers to efficiently complete ID&V requirements and immediately address their actual needs with our frontline staff.
There is a lot of industry knowledge about the quantitative benefits to be achieved in deploying automated ID&V, powered by Gen AI, which provided greater reassurance around expected outcomes.
We began by testing a voice bot using Natural Language Understanding (NLU). While this provided valuable insights into customer interaction behaviours, we discovered an opportunity to improve our verification pass rates. We observed that customers would often continue their conversation flow when prompted for identity information, proceeding with their request before sharing their identity details.
This customer insight led us to explore Large Language Model (LLM) technology, which offered enhanced contextual understanding. The LLM-based voice bot excelled at parsing customer responses and identifying relevant identity information while filtering out conversational context. This advancement resulted in significantly improved pass rates and a more natural customer experience.
This approach closely mimics a human conversation where you would naturally know the difference between the customer giving their identity details and anything that they had said prior to that.
AUTO-ID&V, POWERED BY GEN AI, EMBRACED BY CUSTOMERS AND FRONTLINE STAFF ALIKE
Kristen: How have you measured feedback, both from customers and internally? What kind of response are you getting?
Andrew: We listen to calls and customer response has been overwhelmingly positive. For example, customers have said …
“The experience was great, I loved it”
“The bot experience is easy, it’s seamless.”
Furthermore, we have never seen any negative sentiment about their bot experiences in our CSAT feedback. For me, that stands out because if we were irritating customers, we would hear about it! But there is another perspective which is critical to me. While it’s great that we’ve done this for customers, what I also really care about is that our frontline colleagues enjoy their roles.
The feedback from colleagues has been unanimous and they love that they can get straight in and talk to the customer about what they actually need.
AUTO-ID&V BRINGS RESOUNDING BUSINESS RESULTS
Kristen: Can I ask about the business outcomes and what quantitative results you have achieved? The improvement in pass rate and handling time, for example. Can you take us through what that looked like and how ServisBOT has worked with 1st Central to improve those results?
Andrew: The time savings for the agent equated to about a 5% handle time efficiency and that is significant for us, particularly given the benefit they experienced.
The refinement and iteration process that ServisBOT undertook was impressive. From the initial implementation, ServisBOT’s setup provided us with complete visibility into our funnel. We could immediately identify where customers dropped off, which API calls failed, and when customer utterances weren’t being understood accurately. The reporting provided opportunities for us to fine tune.This transparency allowed our internal teams to enhance the API integration code, resulting in more accurate data retrieval and improved customer validation. As a direct result, we saw a significant increase in our overall pass rate.
Our team collaborated with ServisBOT to review and analyse customer utterances. We upgraded from traditional Natural Language Understanding (NLU-based) to an LLM-based system— implementing a far more sophisticated set of ears. We then methodically optimized the language model’s performance. Through a series of nine targeted LLM fine-tuning iterations, we observed consistent improvements to the pass rate.
After stabilising our initial pass rate, we implemented two additional LLM upgrades that each delivered 2-3% improvements—achieved solely through ServisBOT’s language model fine-tuning and underlying technology upgrades.
This exceptional level of collaboration and ongoing engagement truly distinguishes ServisBOT from other vendors I’ve partnered with. From 1st Central’s perspective, this approach has transformed our automated ID&V process into a remarkable success story.
In terms of our timeline, we started the first Generative AI POC in October 2023 and completed it in February 2024. We went live with the LLM-enhanced automated ID&V bot in August of 2024.
THE PERFECT PARTNERSHIPS WITH OPUS & SERVISBOT
Kristen: Can you tell me how you were initially introduced to ServisBOT.
Andrew: Opus Technologies is our partner for supporting us with our telephony, workforce planning, call recording, and some other technologies and helps us source, procure and implement best of breed technology.
“Opus has a track record of bringing third parties to the table that deliver, that we can trust and that we get along with.”
Opus has been a strategic partner to 1st Central for many years and one of the primary drivers of that is the strength of the relationship. When we engaged with them on our need for automated ID&V they recommended we discuss it with a third party and introduced us to ServisBOT. Opus prides themselves on partnering with technology third parties that are best of breed and ServisBOT sits in that class alongside others.
It was quite clear from the start that ServisBOT knew exactly what they were doing and they were intent on meeting our needs with both the partnership that they provided and the technology that underpinned their service.
Kristen: As you were evaluating how you were going to implement an auto-ID&V voice solution, what were the factors that led you to selecting ServisBOT?
Andrew: ServisBOT has been one of my favorite third parties to work with and there are a few reasons for that.
- Shared Ambition & Pace
Their drive matched ours from day one — and at times, they even outpaced us, pushing the project forward. - Deep Product & Use Case Knowledge
They clearly understood both the technology and its real-world application, helping shape our ideas into impactful solutions. - Collaborative Approach
Rather than just execute, they actively worked with us to refine and evolve the use case for maximum value. - Responsiveness & Adaptability
Throughout the engagement, ServisBOT remained highly responsive and quick to adapt as needs shifted.
“There is an evident capability and competency within everyone I’ve worked with at ServisBOT.”
The other thing that was quite compelling for us was that ServisBOT had relevant UK-based financial services examples of the technology in action. And whilst 1st Central is entrepreneurial, we are pragmatic with our governance and our risk appetite. And knowing that other entities were working with ServisBOT in the same regulated environment as us, doing similar things, gave us a huge amount of assurance.
“ServisBOT has relevant UK-based financial services examples of the technology in action. Knowing that other entities were working with ServisBOT in the same regulated environment as us, doing similar things, gave us a huge amount of assurance.”
“One thing that stood out when we were talking to ServisBOT was it felt like we were building something together over time. And from my perspective, that’s as important as the technology.”
Conclusion
1st Central’s success story demonstrates the value of identifying a clear business problem and strategic use case that can be addressed in a measured and low-risk way utilising Generative AI and LLM technology.
Through its automated ID&V solution, 1st Central achieved impressive results: a 75% pass rate, a 5% reduction in average handle time, strengthened compliance adherence, and overwhelmingly positive feedback from both customers and frontline employees.
The importance of the collaborative approach undertaken with their technology partners cannot be underestimated in this success. By leveraging ServisBOT’s LLM capabilities and Opus Technologies’ support, 1st Central is to rapidly iterate and continuously improve its automated ID&V voice bot.
The journey as told by Andrew offers valuable insights into how financial services organizations can successfully navigate the integration of advanced AI technologies into their business.