Chris Loake, CIO of Hiscox Group, talks about the rollout of Microsoft Copilot, and how to be successful with AI projects.
By the time Chris Loake became the group information officer at Hiscox, in November 2023, the insurance company had already started a journey of piloting using the artificial (AI) intelligence in Microsoft Copilot.
He says, “We’ve had a long-standing relationship with Microsoft.” “When the ChatGPT Revolution began, there were some who thought that this was something we should do.”
Loake says at the time the company didn’t know exactly what Genes would help it achieve. Hiscox decided to roll out M365 Copilot licenses across departments.
He says that “we had a process in which people could apply for these licenses,” which allowed users to buy-in, as they wanted the Microsoft GenAI Tool because they thought it was useful.
AI and IT strategies
Loake, when asked about the impact of AI on the CIO’s role, draws an analogy to the late 1990s. At that time, the web and ecommerce dominated board meetings. “If you look back at the dawn of the Internet, AI is very similar,” he says. “Everyone must have an AI strategy.” It’s the same as having an internet-based strategy.
Loake believes that while many internet-based strategies are not worth their paper, a well-articulated AI-based strategy is like a North Star which outlines a business that is AI-enabled. “We believe AI is a technology that will be the foundation of many, many things,” Says.
This means that Hiscox has a multi-providers environment which offers access to different models, tools and capabilities. Loake says, “We can blend them together as different parts of the organization advance.”
The AI strategy is based on principles of flexibility. “You will have to be flexible and place some bets along the way,” says he. “Your operating model and architecture, as well as your team, must be optimised to allow for agility and the flexibility to pivot when new opportunities arise.”
Another limitation of off-the shelf AI systems is their commoditisation, which means that every business has the same functionality. This can make it difficult to argue that an AI system delivers a business-differentiator.
Loake uses email as an example to show that a commoditised AI is similar to the email application, which every business uses. He asks: “What is my customer journey?” When considering email as a differentiator for business, he says. What do I want to include in email that will offer a tactile or empathetic experience in the customer journey. We believe that the claims process is a crucial moment in an assurance journey. It’s important to be there for our clients and provide a highly empathetic experience. You’re not going to do this over email.”
However, the business has looked at how an AI such as M365 Copilot can support the claims process. This use case differs from how other insurers would use AI – who are more geared toward high volumes and low costs. Loake adds, “They may use it in a totally different way to support the value proposition.”
Takeaways
He says that a number distinct user traits have been identified from the 300-user pilot over a six-month period.
Some people were interested in trying M365 Copilot to solve a particular problem. If it didn’t work out, they would try another solution. This group of users used M365 Copilot a lot, and Loake says: “We saw people who used it frequently also got way more benefit.”
Others were curious about technology and interested in new ways to work, and “kept chipping away” at the software. Loake says that these people tend to self-organise into AI champions. “We created a Microsoft Teams channel for people to discuss Microsoft Copilot,” Says.
The regularity of use and the iterative process to test the Microsoft AI for different problems increased the value of M365 Copilot.
Initial pilots showed that certain job roles could benefit from the tool. Loake says that the tool is especially useful when people need to start from scratch or consolidate large amounts information.
Loake adds that the tool is also useful for people who have to hold many meetings but do not want to attend them all. Meeting summaries and action items from meetings are available as part of M365 Copilot.
Hiscox discovered that some people were in jobs where they didn’t attend many meetings or had to produce discussion documents. They also did not spend much time analysing information from other people. M365 Copilot proved less useful in these roles. Loake says that some people said, “I used it, but it didn’t work for me” and, “It’s rubbish”and they would not use it again.
According to an assessment of the pilot, 15% of those who participated were able to save an extra hour per day in their work. 20% were able to save half an hour; a fifth saw 10-15 minutes; and a quarter reported saving a few minutes.
Hiscox decided to roll out the software to 1,000 users after realizing that some people would be able save significant amounts of time. Loake continued the policy of requiring users to apply for a license, as it increased the likelihood that they would actually use the tool.
Loake found that M365 Copilot champions were more important as people began to use the AI software. This led to the adoption programme. Loake says, “We structured the M365 Copilot champions more and conducted an adoption program where we held drop-in sessions.”
Hiscox appointed M365 Copilot Champions in each department. They were required to help others use the AI tool as part of their jobs. Workshops were organized by groups such as those involved in claims processing.
Now, the firm has rolled out M365 Copilot in 14 countries to over 3,000 staff. Loake says that once you democratise tools like M365 Copilot, some people are going to use them a lot and get a lot of value, while others will only get a limited amount.
For him, making AI tools available to all employees in the company allows them to “feel like they are in the game”and that is something he considers very important. Loake says that you want your users to have the consumer-type tools at their fingertips, and it is important for executives to be aware of what’s going on in their business.
A focused plan
He urges IT leaders not to “go all in with AI” but rather to make some big wagers.
It is not a good idea to have 1,000 proof-of-concept applications that will never be produced. Loake believes it is better to focus on just five applications that can be tested rigorously and then pushed to the business. “You can create these really tangible use-cases,” he says.
Loake says that this allows the IT and business leaders to remain in the game as AI models evolve. As the new models are released and become stable, they can then be used for new, tangible uses cases, using a platform with the flexibility to adapt.
