AI is taking over a growing share of the work in insurance distribution, from preparing quotes to documentation. Yet economic research on “messy jobs” shows why this makes personal advice more valuable: an insurance agent’s worth lies in combining expertise, judgment, and trust. Discover how AI is redefining the role of human agents and what this means for insurers.
Since the arrival of AI chatbots such as ChatGPT, few questions about the future of insurance distribution have been discussed as intensely as this one: Are human insurance agents still needed at all? At first glance, the answer seems clear. AI analyzes risks, compares rates, and explains policy terms and conditions. It now supports even complex advisory conversations surprisingly well. On the face of it, there is much to suggest that face-to-face distribution is losing importance and that AI will replace human agents.
But this conclusion falls short. It rests on the assumption that AI replaces entire professions – an assumption that recent research is calling into question. In reality, AI takes over individual tasks within a field of work. Whether this actually spells the end of a profession depends on how closely those tasks are interconnected.
This is the starting point for research by renowned economists Luis Garicano (London School of Economics) and Jin Li and Yanhui Wu (University of Hong Kong). In their recent paper “Weak Bundle, Strong Bundle: How AI Redraws Job Boundaries” and the book based on it, Messy Jobs, they show that the future of a profession is determined less by individual tasks than by the interplay of expertise, communication, responsibility, judgment, and trust. The economists use the term “messy jobs” to describe professions in which these elements are so closely intertwined that their value cannot be reduced to individual tasks that can be automated in isolation.
For insurers, this perspective opens up a new way of looking at the role of distribution. After all, few kinds of work combine these elements as closely as personal advice.
Individual tasks can be automated – but not necessarily entire professions
Many studies on the impact of generative AI look at individual tasks. As early as 2023, the widely discussed study “GPTs are GPTs” by researchers at OpenAI showed that large language models can support, or even fully take over, a significant share of knowledge-intensive tasks.
Garicano and his co-authors go a step further. Their central argument: companies employ people for their ability to combine different tasks into a functioning whole, not for isolated activities. Analysis, communication, responsibility, coordination, and decision-making are interlinked, and it is precisely this interplay that creates value.
The more closely these tasks are interwoven, the harder it becomes to hand them over entirely to a machine. The authors refer to an “autonomy threshold” – the point AI would have to cross to perform a job independently. This threshold rises wherever the work involves dealing with uncertainty, interacting with people, and taking responsibility.
Insurance distribution is a textbook “messy job”
For insurance distribution, this insight is particularly significant. Of course, many tasks can already be automated today. AI supports needs assessments, prepares quotes, documents advisory conversations, produces risk analyses within seconds, and answers standard questions around the clock. Tasks that took hours or days only a few years ago are now completed in near real time.
Yet this also makes clear where the true value of personal advice lies. The moments that matter most are rarely about missing information; they are about people needing guidance. A business owner reviewing their coverage, for example, may want to know which risks their family or company would actually bear if the worst happened. An affluent private client has to weigh security against returns in their retirement planning. And after a major loss, the focus shifts from contract clauses to priorities, empathy, and the ability to act.
These situations call for experience, judgment, trust, and the ability to bring different interests together. This combination is what makes insurance distribution a profession whose value goes far beyond the sum of its individual tasks.
“AI does not replace insurance agents. It makes them more productive and their core role more demanding.”
The real disruption is unbundling
This leads to one of the most striking findings of the research: AI often does not eliminate professions; it changes what they are made of. Routine tasks are separated out and automated, leaving the activities in which people add the most value. For insurance distribution, this marks a profound change in the agent’s role. Many human agents still devote a considerable part of their working time to research, quote comparisons, documentation, and administrative processes. These are exactly the tasks that AI systems are set to take over.
As a result, the focus of the job shifts fundamentally. Tomorrow’s “human” insurance agents will spend less time explaining products or filling out forms. Instead, they will concentrate on what machines have so far been unable to do convincingly: building trust, facilitating complex decisions, understanding individual priorities, resolving conflicts, and providing guidance to customers, especially in critical situations. So AI does not replace insurance agents. It makes them more productive while making their core role more demanding.
AI creates time for what matters
This development will produce winners and losers. The agents who will come under the most pressure are those whose value has so far lain mainly in providing information or explaining products. AI will deliver exactly these services faster, more cheaply, and at any time of day.
At the same time, there is a major opportunity for advisors who systematically use AI as a productivity lever. Those who prepare quotes in minutes rather than hours, assess risks more precisely, and automate most of their administrative work gain one thing above all: time – time for conversations with existing and prospective customers, time to deepen client relationships, and time for the moments when a salesperson becomes a trusted advisor.
Lasting competitive advantage therefore comes from investing the time saved in better advice and stronger relationships. The most successful agents of the future will look beyond selling policies. They will offer guidance, security, and trust, delivering clear value for customers, which in turn strengthens their business.
The strategic challenge for insurers
This development has far-reaching implications for insurance companies. In the future, competition will depend less on which language model a company uses than on its ability to embed AI consistently in its processes, roles, and decision-making structures. After all, language models are becoming a readily available commodity technology. The real competitive advantage will go to insurers that manage to reshape how people and AI work together, at both an organizational and a process level.
In practice, this means AI helps prepare advisory conversations, needs assessments are largely automated, and documentation takes place almost unnoticed in the background. Underwriting and distribution draw on the same AI systems and work from a shared information base. Insurance agents thus become the orchestrators of an AI-supported advisory process. The true innovation, then, lies not in the language model itself, but in redesigning the business model and the agent’s role.
AI makes good advice more valuable
AI is profoundly changing insurance distribution – though not in the way many expect. The question is not whether AI will replace human agents, but which tasks remain inseparably linked to human judgment, responsibility, and trust.
Current research paints a clear picture. The more complex a profession is, the more tightly its tasks are interwoven, and the more relationships matter, the higher the threshold for full automation. Insurance distribution meets all of these conditions. It is a “messy job” – and that is precisely why AI will not replace the people who work in it.
In insurance distribution, AI will redefine the agent’s role rather than end it. It will relieve agents of routine work, multiply their productivity, and direct their attention to what cannot be automated: building trust, providing guidance, and taking responsibility.
The future therefore belongs neither to digital-only nor to human-only distribution, but to intelligent collaboration between people and machines. This is where the real opportunity for insurers lies. The biggest mistake would be to see AI solely as a cost-cutting tool. Its true potential lies in significantly increasing the value of excellent advice.
“Insurance agents become the orchestrators of an AI-supported advisory process. The true innovation lies not in the language model itself, but in redesigning the business model and the agent’s role.”
Would you like to find out how AI can increase the value of advice in your sales organization? Contact our Simon-Kucher insurance experts.
