Four solutions, one journey: transforming customer experience in motor insurance

Motor insurers manage customer relationships across a complex lifecycle, from an initial policy enquiry and quotation to onboarding, claims and renewal.

Delivering a consistently good experience across these stages is difficult. Customer information may be distributed across policy documents, partner platforms and legacy systems, requiring colleagues to search for information and move between multiple applications while speaking to the customer.

At the same time, customers expect clear answers, competitive products and prompt support, particularly when making a claim which is a sensitive time in anyones life.

Working with our client, we identified four points in the customer journey where AI assistance and process automation could reduce manual effort, improve service and deliver measurable business outcomes:

• Policy support
• Quotes and purchases
• Customer onboarding
• Claims and renewals

Rather than applying the same technology to every problem, we designed each solution around the needs of the customer, the colleague and the process.

Making policy information easier to understand

Insurance policies contain two broad categories of information.

The first is the standard wording that explains what a policy covers, its exclusions and the obligations of both the insurer and the customer. These documents are often shared by customers with the same type of cover, but their length and technical language can make it difficult to find and understand specific information. In particular, customers for whom English is their second language would often find it challenging to understand the details of their coverage.

The second category is personal to the individual customer. It includes their selected level of cover, insured vehicle, named drivers, optional benefits, promotional terms and documentation from relevant partners, such as breakdown providers.

We deployed a customer-facing AI assistant that could search both types of information and respond to questions about an individual policy. Customers could ask what their cover included, explore hypothetical situations and locate relevant terms without needing to telephone the contact centre.

The assistant combined retrieval-augmented generation with context-aware document search. It identified relevant information from approved policy documents and used it to produce a clear response.

Access to personal information took place within authenticated sessions. Customer details and individual policy documents were handled in isolated contexts, supported by appropriate access controls, to keep one customer’s information separate from another’s.

The service was designed to complement existing support channels. When a question was ambiguous, required judgment or fell outside the available information, the customer could be directed to a colleague for further assistance.

Helping colleagues produce quotes more efficiently

Price-conscious customers increasingly compare insurers across multiple channels before making a purchase. Some begin through a price-comparison website, while others contact an insurer directly to discuss their requirements and available options.

For customer service colleagues, producing a suitable quote can involve gathering information from the customer, checking existing records, retrieving vehicle details and entering data into a separate quoting platform. Legacy systems can make it time-consuming to compare different combinations of cover, excesses and optional benefits during a live conversation.

We developed a colleague-facing AI assistant to coordinate this process.

When a customer called, the system used their telephone number to search the customer relationship management platform. If a matching record existed, it retrieved the relevant information for the colleague. If the customer was new, it initiated a structured process to collect the required details without asking unnecessary questions.

The customer could provide the vehicle registration number, allowing the assistant to retrieve available vehicle information and coordinate the necessary checks. Where appropriate, external market data could provide supporting information about the vehicle’s value, subject to the insurer’s approved valuation and underwriting processes.

The assistant then guided the colleague through the required questions and documentation. Once the necessary information had been collected, it interacted with the existing quotation system to generate eligible options within the insurer’s approved rules.

Instead of repeatedly entering and adjusting information manually, the colleague could compare relevant combinations of cover, excess and optional benefits in real time. They could then explain those options to the customer, answer questions and help them make an informed choice.

Once the customer selected an option, the assistant supported the colleague in confirming the cover and completing the purchase.

The technology accelerated the process, but it did not replace the controls governing product eligibility, pricing or underwriting. Decisions remained subject to the insurer’s established rules, with colleagues involved wherever explanation or judgment was required.

Automating customer onboarding

After purchasing a policy, the customer needs to be onboarded and receive the correct documentation.

Unlike policy enquiries or claims, onboarding is generally a more predictable and linear process. The required data is known, the systems of record are defined and the sequence of steps is repeatable.

This made process automation more appropriate than a conversational AI agent.

We built an automation layer on top of robotic process automation technology to transfer validated information between systems, populate the required fields and generate policy documents. The solution was able to work with existing applications, including legacy platforms that did not provide modern integration options.

Validation checks were applied throughout the workflow. Cases that met the expected conditions could be processed without routine manual handling, while incomplete records, conflicting information and system exceptions were routed to a colleague.

This reduced repetitive administration and accelerated the issuing of policy documents. It also allowed colleagues to focus on the smaller number of cases that genuinely required investigation or intervention.

Supporting more empathetic claims conversations

Claims are very different from onboarding. Customers may be distressed, their vehicle may be unusable and their daily life may have been disrupted.

In these moments, the quality of the human conversation matters. However, claims handlers often need to divide their attention between the customer and multiple systems. They need to check policy details, validate information, review documents and confirm services available from external partners, all whilst speaking to a likely distressed customer.

We created a colleague-assisted claims solution with a dedicated interface and specialised AI capabilities operating behind it.

As the customer provided information, the solution assembled the relevant context for the claim. Specialised components could verify policy details, organise information about the incident, identify the applicable level of cover and retrieve information from suppliers or partners. For example, the agent could both validate whether the customer was eligible for a courtesy vehicle, and provide short lists of nearby providers to the colleague based upon the customers location.

This gave the claims handler a clearer view of the information needed to support the customer. Instead of repeatedly searching documents and switching between systems, they could focus more fully on the conversation, explain the next steps and respond with empathy.

The system supported information retrieval and verification, while claims decisions and exceptional circumstances remained subject to the insurer’s established processes and appropriate human judgment.

The same colleague-assisted approach also supported renewal conversations by bringing together relevant policy information and available options. This helped colleagues explain changes clearly and guide customers through the renewal process without manually assembling information from several systems. The quotation workflows developed for new customers were also reused for renewals, reducing duplication and making the solution easier and faster to scale.

Delivering measurable outcomes

Together, the four solutions introduced targeted automation across the customer lifecycle without treating every process as the same kind of problem.

Customer-facing AI made policy information easier to access. Colleague-facing assistance reduced the effort involved in quotations, claims and renewals. Process automation accelerated predictable onboarding tasks.

Following deployment, average handling time across the measured quotation and claims journeys fell by 30%. Net Promoter Score among customers using the telephone channel increased by 14% over the evaluation period.

These results reflected more than a reduction in administration. By making relevant information available at the point of need, the solutions gave colleagues more time to listen, explain and support the customer.

The four solutions were designed and deployed in just eight weeks.

Applying the right technology to each problem

One of the most important lessons from the engagement was that successful automation does not mean deploying an autonomous AI agent everywhere.

Different moments in the customer journey require different approaches:

• Customers looking for policy information benefit from conversational access to trusted documents.
• Colleagues producing quotes benefit from guided workflows and controlled system integration.
• Predictable onboarding processes are better suited to deterministic automation.
• Claims conversations benefit from AI that supports human empathy and judgment rather than attempting to replace them.

In each case, the design was shaped by the process, the available information and the consequences of getting a decision wrong. Access controls, validation, exception handling and human escalation were treated as part of the solution rather than added afterwards.

The result was a connected set of capabilities designed to improve both operational performance and the customer experience.

To learn more about this case study, or how AI agents and automation could transform your customer journeys, get in touch.