Firemind had the pleasure of co-hosting a packed insurance roundtable in London with Amazon Web Services (AWS). Together with guest speakers from Convex and AWS, and industry leaders from across the specialty and commercial market, we explored a question dominating the sector right now:
Where is AI delivering real operational value in insurance today, and what’s still holding the industry back?
Below, we highlight some of the themes from the evening and what they mean for the future of underwriting, claims, and operations.
AI is removing the friction that slows down underwriting
One of the strongest messages came from Andrea Moscatelli, who walked us through how Convex is using AI to scale underwriting productivity without increasing headcount.
While the industry often talks about AI’s long-term potential, Convex is already applying it to high-value, immediate bottlenecks, especially the manual admin that consumes underwriters’ time.
Practical use cases in production today
- Automated submission summaries: AI condenses hundreds of pages into clean, structured insight packs in minutes, allowing underwriters to review far more risks with the same resources.
- Classification and data extraction at scale: Millions of data points from emails, PDFs and attachments are classified and enriched automatically.
- Faster triage and risk ranking: AI highlights the most relevant risks first, giving teams a head start on the opportunities most aligned with appetite.
The takeaway: AI doesn’t need to replace underwriters to generate impact. It simply needs to remove the repetitive work blocking their capacity.
AWS demonstrated how a modern underwriting stack works
While Convex showcased the “what,” AWS demonstrated the “how”: showing live how a modern underwriting workflow becomes significantly faster, more consistent, and more transparent with AI.
Key capabilities highlighted
- Intelligent Document Processing (IDP) to extract, classify and structure unstructured submission data
- External data enrichment across web sources, financials, D&B, and OSINT
- Explainable appetite scoring showing why a risk is ranked high or low
- Agentic workflows that handle multi-step tasks such as memo drafting or gap analysis
- Flexible interfaces that fit directly into an underwriter’s existing environment
In real deployments, AWS has seen carriers accelerate underwriting cycles dramatically, in some cases enabling same-day decisions for complex commercial risks.
Firemind’s perspective: AI must solve real operational pain, not just theoretical use cases
During the discussion, Ahmed Nuaman from Firemind emphasised a critical but often overlooked point: AI adoption in insurance accelerates only when it tackles real, everyday friction, not abstract innovation themes. Drawing from current client work, he highlighted the significant, often untapped opportunity within claims and assistance operations, where decisions are complex, time-sensitive and emotionally charged. Many of these workflows: medical triage, repatriation decisions, hospital cost prediction, case strategy and high-volume claim handling, still rely on fragmented data, manual interpretation and multiple handoffs. Assistance is often overlooked because it sits at the intersection of claims, operations and medical decision-making, which means responsibility is shared, but ownership is rarely clear. Ahmed explained that AI can play a transformative role by structuring unstructured medical information, surfacing contextual insights and supporting case managers with more consistent, informed decision-making. Crucially, he emphasised that the insurers moving fastest are not waiting for perfect data or perfect systems; they are focusing on targeted, workflow-level improvements that remove cognitive load, elevate decision quality and support teams at the moments that matter most.
Trust and explainability are now core requirements
A recurring theme from both speakers, and echoed across the room, was this: AI adoption isn’t a technology problem. It’s a trust problem. Underwriters, claims handlers, and operational teams want to understand the reasoning behind an AI recommendation, not just receive an output.
The features that build trust include:
- Showing where every data point was sourced
- Transparent scoring logic
- Human-in-the-loop workflows
- Clear audit trails
- Model guardrails to prevent hallucination
Solutions that offer this level of transparency see adoption accelerate significantly faster.
The appetite for AI is high but governance is the new bottleneck
Nearly every organisation in the room is experimenting with AI. But when it comes to getting solutions into production, many face the same hurdles:
- Complex legacy systems with limited integration points
- Newly formed AI governance boards adding review cycles
- Regulatory uncertainty around automated decisions
- Cultural resistance among operational teams
- Data quality and data access challenges
The result? Lots of promising POCs but only a small number of live, scaled deployments.
This is where insurers are actively seeking specialist partners who can help them de-risk implementation while accelerating time-to-value.
The smartest insurers are starting small and scaling quickly
Rather than chasing large, multi-year transformations, the market leaders are focusing on targeted, high-impact workflows such as:
- Document summarisation
- Submission triage
- Risk memo generation
- Claims document analysis
- Low-level operational automation
These deliver fast wins, build internal confidence, and open the door to broader AI strategies.
The next step is about real value, not experiments
If there was one shared sentiment across the room, it was this: The insurance industry is no longer asking whether AI will transform underwriting and claims but how quickly and safely they can adopt it.
The carriers moving fastest are those who:
- Choose practical, explainable solutions
- Invest in governance and guardrails
- Engage underwriters early
- Focus on workflows, not just models
- Partner with specialists who understand the complexity of insurance operations
This is where Firemind is proud to support the market helping insurers move from experimentation to real, governed impact at speed.
Looking ahead: AI leap for Insurance
A huge thank you to our speakers, for sharing their perspectives and pushing the industry conversation forward.
And thank you to everyone who joined us. The energy in the room made one thing very clear:
Insurance is ready for its next leap, and AI will be the catalyst.
If you’d like to explore the use cases discussed, see a demo, or join our next roundtable, our team would love to connect.
Take the next step
Discover solutions, customer success stories, and events on our dedicated Insurance page.
