Talent [R]evolution

AI in insurance is game-changing – but you need the right people

Reading Time: 7 minutes

As insurance professionals, we find ourselves in the landscape of Insurance 2.0, where artificial intelligence (AI) takes centre stage. AI in insurance is set to revolutionise the industry, creating both opportunities and challenges. With so many possibilities on the horizon, examining the potential use cases is a fruitful field of discussion.

As a natural-born scientist with a PhD in physics, I’ve always been tantalised by the possibilities presented by computing and technology. And, as somewhat of a technology and digitization veteran, I’m excited to see what this new era will bring. I spent several years in R&D at Siemens in Munich, as a consultant at Booz Allen Hamilton and served a decade with Deutsche Telekom, ultimately as the Vice President for B2B strategy in Germany.

After that, I transitioned into an independent advisory role, focusing on technology driven growth in the tech, telco, banking and insurance sectors. I’m an active member of the Outvise network and it was a pleasure to be invited to talk about the new vanguard of AI in insurance with a distinguished panel of experts, including Denys Holovatyi, Chief Artificial Intelligence Officer at OSNOVA; Dr. Jesper Heide, AI expert & CTO for several insurance, reinsurance, and consulting companies; and Pau Cerdà, Co-founder of Outvise. 

During our hour-long discussion and Q&A, each panellist shared authentic use cases and provided valuable insights into navigating the dynamic intersection of technology and insurance. We also explored the skills, resources and profiles essential for success in an AI-driven landscape, including how to build teams tailored for the insurance domain. 

For all the insights, check out the webinar. Or, if you’ve only got a few minutes, I’ll run you through my take-aways here.

How can insurance companies use generative AI?

If you check Outvise’s YouTube channel, you’ll notice that three years ago, several discussions were initiated on AI, primarily focusing on supervised learning. Back then, despite exploring various possibilities and pilot cases, the attention from C-levels was somewhat limited. 

Fast forward to last year, specifically on 30th November, ChatGPT was launched by OpenAI. This drew significant attention and changed the game. CEOs and C-level executives grasped the unique qualities of AI, especially in terms of potential savings and efficiencies. Unlike traditional digital transformation projects with intricate processes and business cases, AI now presents a clear and quick demonstration of value. Prototyping is key to quickly implement selected use cases and demonstrate their success. Furthermore, it is equally important and helpful to use prototyping to identify which use cases can be successfully implemented and which approaches are less promising. With all approaches it is always important to keep a business focus and implement use cases with real added value for the company.

AI in insurance could yield particular results. This is due to the industry’s complexity, involving a mix of human and tech interactions, complex workflows and heavy regulation. Take claims management; the process begins with a customer initiating a claim through a call to the company. A designated customer service representative will record accident details and navigate through various associated tasks. 

Next, the Claims Examiner carefully checks all the submitted documents to make sure everything needed is there. A broker or agent will help the policyholder communicate with the insurance company, clarifying policy details. Then, the Claims Adjuster takes charge, investigating the claim’s details and negotiating settlements. In tandem, the Claims Appraiser assesses the value of the damaged property. Last but not least, the Subject Matter Expert (SME) contributes their specialised knowledge to ensure accurate assessments and decisions. 

It is clear, there are myriad roles at every stage in the process, each of whom has to follow meticulous processes. Indeed, wherever a meticulous process is involved, AI can do a lot to aid speed and precision. For example, how is AI transforming insurance underwriting? Before the process described above even starts, the Underwriter assesses risks and sets policy terms. With predictive analytics, the Underwriter can gain deeper insights into risks, while real-time data enables on-the-fly quotes. This dynamic approach allows for personalised policies, tailored coverage and lower premiums. Not only does AI in insurance enhance customisation, but it’s also a driving force behind cost reduction, ultimately delivering savings to customers.

As such, it will be essential to work closely with the entire internal team to identify use cases relevant to your organisation and your processes, whether it be in claims management, marketing, or IT strategy. Following close collaboration with this team, you will need to engage another – those who will prototype the AI model for your given use case.

ai in insurance use cases
AI can significantly enhance speed and precision in any meticulous process it is applied to.

Building an AI development team

The AI development process can be distilled into four straightforward steps: first, data preparation; second, selecting a suitable model; third, validating the model through fine-tuning; fourth, deploying it, whether in an app, the cloud, or the internet. Bear in mind that playing around with small pilots might not yield lasting support or budgetary commitment. Scalability is another critical factor; solutions should be designed with scalability in mind, moving beyond mere pilot projects. 

Assembling the right team is challenging, necessitating a blend of internal and external expertise to foster learning and knowledge retention. Integration teams must include external partners, ensuring knowledge transfer and internalisation of key insights. These steps entail acquiring new and specific knowledge, which is often lacking in many companies. Seeking external expertise is therefore crucial.

As in the insurance industry itself, the development process requires a team where each member has a distinct function. Perhaps the most important team member will be provided by the client; a Subject Matter Expert. This is because it’s crucial to ensure that the addressed use cases are highly relevant and provide substantial business impact from the outset. This person will be immersed in the daily business process, possessing insights into the tasks performed regularly. They’ll be crucial to analysing the use cases of AI in insurance, alongside the technical team. 

For technical roles, the setup varies based on the use case. For instance, extracting data from systems like Guidewire requires expertise in the respective data model. Here, you might need a professional familiar with the intricacies of Guidewire’s data structure. On the data-related front, an AI Engineer plays a pivotal role. This is a broad role that evolves based on the project’s needs – sometimes focusing on machine learning research and, more often, on applied tasks using common libraries like Long Chain or vector databases such as Pine Cone.

There’s also a distinct role – the Prompt Engineer. This individual, often a former business analyst, crafts complex prompts and prompt sequences. Crafting a prompt is crucial as it’s the means of communication with the language model. While straightforward prompts work well for personal use cases, in the corporate environment with diverse data sources, complex processes and legal constraints, prompt engineering becomes an art and science. The AI Engineer builds the architecture, defining how prompts are sent to the API, and how outputs are processed, while the Prompt Engineer crafts and experiments with the wording to ensure precise and reliable responses.

Classic roles like Cloud Engineers and Data Engineers are also necessary for infrastructure setup in any data project, including generative AI projects. The challenge lies in making language models more reliable by structuring prompts effectively.

Beyond the nuts and bolts, legal and regulatory considerations, data security, and governance should be addressed proactively when it comes to AI in insurance. These discussions, especially within larger corporations, can be prolonged and require dedicated efforts from knowledgeable individuals. In addition to the profiles already mentioned, a seasoned Program Manager with a strategic outlook is vital to guide teams effectively. Managing legal intricacies can be as demanding if not more so, than the actual IT project, emphasising the need for robust program management.

Unraveling the talent landscape for AI in insurance

Outvise has various long-standing clients in the insurance industry, including AXA Allianz, Zurich and Ergo. In the past, Outvise has provided Data Experts, Machine Learning Engineers, Big Data Architects, and BI Consultants among others. However, the landscape becomes more challenging when it comes to generative AI. The profiles required for insurance in AI are relatively new, making them harder to find, especially given the limited experience in the market. Meanwhile, demand for such profiles has gone up by 30%, according to data from the recruitment team.

Prompt engineering, for example, is a fascinating role. There are questions as to whether it’s more of a skill set or an independent job; the nature of prompt engineering is such that individuals within companies naturally become prompt engineers as they integrate generative AI into their daily tasks. In this sense, prompt engineering can be considered a skill. 

However, when we look at it as a distinct job role, the scenario changes. On LinkedIn, the number of job postings for Prompt Engineers increased from 100 to around 500 since June. Yet, finding professionals specifically describing themselves as Prompt Engineers is still rare. Naturally, this scarcity poses challenges for AI in insurance

Despite the challenges, the increasing number of individuals in this space is promising. I’d caution against expecting extensive experience, as AI in insurance is a rapidly evolving field. Even a professional with two solid projects in the last year can be considered well-versed in this domain. It’s not about a decade of experience in a technology that’s only been around for five years.

This is where platforms like Outvise come in so useful. When putting together a team, it’s crucial to cast a broad net rather than narrowing it down too soon. In my experience, being too narrow in your search increases the likelihood of missing out on skilled individuals. Specificity is essential, but an overly narrow focus can result in overlooking talented people. Examining CVs and competencies comprehensively allows you to assemble the right team. 

Companies may sometimes demand specific technical expertise, but not all tech elements are differentiating factors. Some technical skills can be acquired and filtering too narrowly might lead you to miss out. Turning to a curated yet extensive platform like Outvise can help you access a broad range of experiences while ensuring quality.

Set the strategic foundations for a successful project

An AI project begins with a clear vision of the desired outcome, a well-defined use case, and a thorough understanding of the potential business impact. Identifying where to allocate the team, whether in sales, marketing, or CIO areas, is essential. Clarifying the working mode and decision-making levels is the next step. I emphasise the importance of doing this preliminary work before diving into the novel world of AI in insurance. It’s crucial to establish ownership, organisational structure and support upfront.

Then comes putting together the perfect team. Outvise operates as a network of Business Tech freelancers. Our community spans over 40,000 tech freelancers globally with a focus on high-end, highly skilled tech freelancers. What sets Outvise apart is the platform approach; unlike traditional agencies with a limited pool of resources, Outvise leverages its extensive network to cherry-pick the best profiles for each opportunity.

This technology-driven approach enhances our community and allows us to bring in top experts, such as the AI Specialists who joined me on this panel. If you’re looking for an expert to provide you with initial consultancy or full project delivery, you can post a talent request for free.

Marc is a senior advisor specializing in strategy and growth for the technology and telecommunications sectors. With expertise in commercial Post-Merger Integration (PMI), strategic business reviews, reorganization for efficiency, and digitalization of sales processes, Marc has led initiatives for prominent companies including a European fiber carrier and international mobile operators. Formerly, he served as Vice President of B2B Strategy at Deutsche Telekom.

No comments yet

There are no comments on this post yet.