Will AI Underwrite the Next Decade?
- Dani Elliott-Booth
- 6 days ago
- 4 min read
The recent launch of CFC's Lane Assist gave the conversation at Camelot SPARK an unusually concrete starting point. Rather than debating whether AI might handle underwriting submissions in the future, our speakers were discussing a system that’s already doing it.
That specificity changed the tone of the debate from the very beginning. This wasn't about the future. It was about now, and what the industry does with it.
This was a punchy, candid discussion between technology builders and underwriting practitioners. The session explored not just what is possible, but what is responsible, sustainable and human.
The conversation was open and grounded in lived experience. What follows are the key learnings the room took away.
From concept to catalyst: why Lane Assist changed the conversation for AI underwriting
AI in underwriting is not new. What is new is seeing agentic capability move from pilot decks into live production, handling real submissions at speed.
Lane Assist was not presented as a moonshot. It was framed as a system designed to remove friction from the underwriting process: ingesting broker emails, extracting intent, applying underwriting rules, configuring terms, pricing risk and returning a quote, all while keeping a human underwriter firmly in the loop.
That reality changed the tone of the debate immediately.
This was no longer about whether AI might change underwriting one day. It was about how it already is, and how organisations choose to respond.
What gets automated and why that matters
A clear line emerged early in the discussion: automation works best around judgement, not instead of it.
The strongest use cases described were those targeting the “donkey work” of underwriting: re‑keying data, checking completeness, applying standardised rules and producing draft outputs. These are the tasks that consume time without creating differentiation.
By automating these steps, AI creates three tangible outcomes:
Faster response times for brokers
Lower cost per submission
Increased underwriting capacity without proportional headcount growth
Crucially, automation here is not about speed for its own sake. It is about freeing underwriters to spend more time where human expertise adds the most value: complex risks, broker conversations, portfolio thinking and innovation.
What doesn’t get automated: judgement, accountability and trust
This is just as important was what doesn’t get automated.
Across the panel, there was strong agreement that judgement remains human. This isn’t because technology isn’t improving, but because trust, accountability and responsibility still sit with people.
Human‑in‑the‑loop design emerged as a defining principle. Underwriters can see:
What changed from default
Which rules were applied
How pricing was derived
Where assumptions sit
That transparency matters. It builds internal confidence, accelerates regulatory comfort and prevents the slow erosion of professional judgement into “computer says yes”.
AI, in this framing, is not an underwriter. It is an accelerator.
Governance: the real competitive advantage
If Lane Assist was the catalyst, governance was the backbone of the conversation.
The debate repeatedly returned to the same uncomfortable truth: AI capability is becoming accessible to everyone. What will differentiate organisations is not the model they choose, but how well they govern it.
The harder question, and the one the session kept returning to, is how organisations maintain genuine oversight as these systems evolve. Trusting a model enough to step back is not a decision you make once. It's a discipline that has to be built, tested and revisited as capability grows and stakes increase.
There were no easy answers, and that was the point. The session reinforced that successful AI adoption is not a one‑off launch, but an iterative, organisation‑wide discipline, requiring underwriting, actuarial, compliance, technology and leadership to move together.
Systemic risk and the danger of comfortable correlation
One of the most thoughtful moments came when the discussion turned outward, from firm‑level benefit to industry‑level risk.
As more carriers deploy similar agentic approaches, trained on overlapping data and mathematical structures, there is a risk of underwriting decisions becoming overly correlated. Human error diversifies risk; algorithmic error, if unchecked, can amplify it.
The counterbalance was equally important. AI does not invent appetite — it enforces it. Differentiation still lives in underwriting guides, proprietary data, governance standards and human calibration.
The takeaway focused on responsibility. Systemic risk is not a reason to stop, but it is a reason to think together as an industry.
Brokers, outcomes and uncomfortable truths
From a broker perspective, the message was refreshingly direct: outcomes matter more than methods.
Faster quotes, consistent decisions and clearer communication are what brokers value. In many cases, they will not know whether AI was involved. And they don’t need to.
That said, the session did not shy away from uncomfortable truths. Increased consistency removes the opportunity for “gaming the system”. It also raises expectations. As agentic underwriting becomes visible, scrutiny will increase. And rightly so.
The underwriter of the future: fewer keystrokes, broader responsibility
Perhaps the most human part of the debate focused on talent.
If junior underwriters no longer learn by manually processing thousands of simple risks, how do they develop instinct? How do we actually train the next generation of exceptional underwriters?
The underwriter of the future looks less like a transaction processor and more like a portfolio manager: calibrating appetite, interrogating models, reviewing outcomes and shaping frameworks. Training will evolve and skill sets will shift as a result, but excellence remains the goal.
AI is not good at genuinely novel thought, whereas people still are.
Why Camelot conversations matter
This debate captured exactly why the Camelot Network exists.
Progress comes from honest conversations between people willing to share what they are building, what they are worried about and what they are learning in real time.
The room explored AI not as hype, but as responsibility. That is how trust is built and how industries move forward.
Continue the conversation at Camelot SHIFT
If this discussion resonated, (if you are navigating AI, underwriting transformation, governance or talent within your organisation) we would love you to join the next conversation in September.
Camelot SHIFT brings senior insurance professionals together to explore what’s changing, what matters and how we walk forward, with confidence, as a community.
