AI in Underwriting Report 2025
Crucial Insights on the risks and benefits of using AI in underwriting
Intelligent Insurer surveyed over 200 insurers and brokers to find out how AI is impacting underwriting in 2025.
Here is what they told us...
Change is constant in insurance, but the AI revolution in underwriting isn't just another shift - it's a tectonic one. What was once a distant possibility is now a front-line reality. This year's AI in Underwriting Survey makes it clear. AI is no longer about "if", but about how fast and how deeply it will reshape underwriting.
At Send, we're proud to sponsor this report because it captures the industry's pivot from dabbling to doing. What were once isolated experiments are now evolving into enterprise-wide strategies. AI is no longer an accessory—it's becoming the backbone of underwriting.
The numbers say it all. Last year, almost half of the underwriters we surveyed (48%) hadn't touched AI, today that's down to 19%. The rest are either piloting, deploying, or scaling AI, and the momentum isn't slowing. AI has graduated from "nice to have" to strategic imperative, with 77% of respondents betting it will be the biggest lever to boost underwriting performance and profitability in the year ahead and 87% believing that successful deployment will create a lasting competitive advantage.
Augmentation, not replacement
But here's the kicker: it's not about replacing people. AI can act as the bridge that marries the art and the science of underwriting, helping underwriters to use their talents more efficiently. Over half of those surveyed this year now see AI as a co-pilot, amplifying human judgement by taking on the heavy data lifting and routine work. The conversation has evolved: AI isn't a solo act—it's the partner that enables underwriters to focus on what machines can't do—creativity, nuance, and risk intuition.
Investment follows intention. Seventy five percent expect AI investment to grow over the next 1–2 years, shifting from pilot to enterprise-grade deployments. This is about building AI into the very DNA of underwriting, across products, teams, and workflows.
While automation and data crunching remain front and center, new use cases—like real-time underwriting and emerging risk detection—are gaining ground. The rise of decision-support tools signals a maturing mindset: AI is here to enhance expertise, not just accelerate it.
Looking ahead, generative AI and agentic AI are already moving from buzzwords to boardroom agendas. More than 63% of respondents see generative AI impacting underwriting soon, with a fifth already experimenting with AI copilots that can independently execute tasks. This is the next frontier, promising to shift AI from a static tool to a dynamic teammate.
The next move is execution
The challenges have moved from theoretical debates to practical hurdles: navigating AI regulations, integrating with legacy systems, and maintaining security remain top of mind. Thirty percent of respondents still lack confidence in their ability to manage AI-related regulations, highlighting a clear gap between ambition and operational readiness. However, concerns around ethics and cost are easing as experience builds confidence.
Importantly, AI is emerging as a talent magnet. Sixty percent of insurers believe AI adoption helps attract and keep top underwriting talent by making roles more engaging and strategic—an advantage no company can afford to ignore in a tight market.
AI adoption varies by sector, with data-rich lines like Motor and Commercial Property leading the way, while more complex and heavily regulated lines take a more cautious approach. Yet across the board, 78% expect AI to be embedded in day-to-day underwriting within three years—many even sooner.
The bottom line? AI isn’t a replacement for underwriting’s human core. It’s the tool that empowers judgement, sharpens insight, and unlocks capacity. The future belongs to insurers who treat AI as a strategic partner, not a magic bullet.
At Send, we see AI as a foundational shift—one that will reshape how underwriters work, how insurers compete, and how the market evolves. The time to move beyond pilots is now. AI isn’t waiting in the wings; it’s already on the underwriting floor.
-Andy Moss | Co-founder, CEO, Send
Questions and Answers
Question 1: AI presents the biggest opportunity for improving underwriting performance and profitability in the next 12 months
Answer: 77% of respondents said yes, 23% said no.
Compared to 2024, A strong 77% of respondents agree that AI presents the biggest opportunity to improve underwriting performance and profitability over the next 12 months, a notable rise from 65% in 2024
Question 2: In your opinion, will insurers that successfully deploy AI in underwriting have a stronger competitive advantage?
Answer: 87% of respondents said yes
Question 3: What best describes your current use of AI in underwriting?
Answer:
| Current AI Use | % of Respondents |
|---|---|
| Exploring | 34% |
| Using in specific lines | 20% |
| Piloting | 19% |
| Not using | 19% |
| Scaled across the business | 8% |
Adoption of AI in underwriting has moved decisively forward; in 2024, 48% of respondents were not using AI in underwriting. In 2025, only 19% were not using AI—with the rest now actively exploring, piloting, or deploying AI. A clear sign the industry is moving from curiosity to action.
Question 4: How do you see AI impacting the role of underwriters over the next 2–5 years, and what best describes your organization’s approach to AI in the next 12 months?
Answer:
| AI Impact & Organization's Approach | % of Respondents |
|---|---|
| AI will moderately change the role, and we're beginning to explore its use strategically. | 49% |
| AI will significantly change the underwriter's role, and it will be a key part of our strategy over the next 12 months. | 27% |
| The role will stay largely the same, but we are experimenting with AI in limited areas. | 18% |
| Not sure yet / Still evaluating. | 4% |
| We don't see AI having a major impact and are not prioritizing it in the near term. | 2% |
Question 5: Which underwriting functions do you currently apply or are considering applying for?
Answer:
| Underwriting Function | % of Respondents |
|---|---|
| Document ingestion (e.g. unstructured documents) | 61% |
| Risk assessment | 60% |
| Pricing and rating | 42% |
| Fraud detection | 40% |
| Submission triage | 33% |
| Portfolio management | 28% |
| Not currently using | 12% |
Other functions or considerations for AI include:
- Process optimisation: SOP creation, preventative maintenance, and operational efficiency.
- Business functions: Applied in HR, sales, marketing, finance, and business intelligence.
- Support functions: Contract review, reserving, quality assurance, and control.
- Product and programme development: Supporting innovation, data organisation, and presentation.
- Advisory and training: Used for second opinions, coaching, and consulting.
Question 6: What impact has AI predictive analytics, automation, machine learning) had on your underwriting outcomes so far?
Answer:
| Underwriting Outcome | % of Respondents |
|---|---|
| Too early to tell | 40% |
| Reduced manual workload | 30% |
| Not applicable | 17% |
| Faster quote turnaround | 6% |
| More accurate pricing | 5% |
| Improved loss ratio | 2% |
Question 7: To date, what impact—if any—has Generative AI (e.g., ChatGPT, LLM-based tools) had on your underwriting workflows or decision-making?
Answer:
| Impact of Generative AI | % of Respondents |
|---|---|
| A somewhat positive impact | 38% |
| A significant positive impact | 25% |
| Only a small positive impact | 22% |
| No impact at all | 15% |
Question 8: To what extent is your underwriting department exploring or using Agentic AI (e.g. AI-powered copilots that independently plan, execute and refine underwriting or pricing tasks)?
Answer:
| Agentic AI Adoption | % of Respondents |
|---|---|
| Exploring the potential but haven’t implemented anything yet | 34% |
| We are not familiar with Agentic AI | 27% |
| Aware of Agentic AI but haven’t considered it yet | 18% |
| Piloting or testing Agentic AI capabilities | 14% |
| Actively using Agentic AI in underwriting | 7% |
Question 9: What do you currently value—or would most value—from a service provider offering AI underwriting tools?
Answer:
| Feature or Capability | % of Respondents |
|---|---|
| Easy integration with existing systems | 36% |
| Proven use cases | 27% |
| Model explainability | 11% |
| Regulatory compliance support | 9% |
| Customisation/flexibility | 7% |
| Training and change management support | 6% |
| Transparent pricing | 4% |
Question 10: How do you expect your company's investment in AI for underwriting to change over the next 12–24 months?
Answer:
| Expected Investment Change | % of Respondents |
|---|---|
| It will grow significantly | 40% |
| It will grow slightly | 35% |
| Unsure | 9% |
| Not currently investing in AI for underwriting | 8% |
| It will stay the same | 6% |
Investment in AI: 2024 vs. 2025
The market is maturing. What was once cautious exploration is now shifting toward bolder AI investment strategies, with more insurers moving from pilot to scale in the next 12–24 months.
There’s a clear shift toward greater commitment and confidence:
- Significant investment growth increased by 10 percentage points, rising from 30% in 2024 to 40% in 2025, indicating stronger intent to scale AI initiatives.
- "Stay the same" responses declined sharply, falling from 20% to 6%, reinforcing the momentum behind active investment.
- New in 2025: 9% of respondents reported being unsure, highlighting the need for stronger business cases and clearer outcomes to support AI adoption.
Question 11: Do you understand the regulatory landscape surrounding AI, and what is required to implement AI responsibly?
Answer:
| Level of Understanding | % of Respondents |
|---|---|
| Somewhat understand | 52% |
| Struggling to understand | 18% |
| Completely Understand | 17% |
| Do not understand | 13% |
Takeaway
While most respondents still lack full confidence, the year-on-year comparison between 2024 and 2025 suggests the industry is steadily gaining a better understanding of AI regulation. However, with nearly one-third still struggling or unsure about the regulatory landscape, there remains a clear need for greater education, clearer guidance, and stronger support to ensure responsible AI implementation at scale.
Question 12: Insurers embracing AI will help attract and retain top underwriting talent
Answer:
| Response | % of Respondents |
|---|---|
| Agree | 60% |
| Not sure | 32% |
| Disagree | 8% |
Question 13: What are the benefits of using AI in underwriting? (Top 5 were selected)
Answer:
| Benefit | % of Respondents |
|---|---|
| Extracting and analysing unstructured data | 67% |
| Automating and streamlining underwriting tasks | 62% |
| Faster decision-making and speed-to-market | 53% |
| Helping with risk triage and selection | 38% |
| Early identification of emerging or hidden risk factors | 28% |
| Managing complex data queries | 27% |
| Improved portfolio insights and profitability tracking | 26% |
| Increasing capacity in underwriting teams | 25% |
| Enriching data | 24% |
| Supporting real-time underwriting | 20% |
| More consistent, accurate, and reliable decisions | 19% |
| Smarter renewals and cross-sell opportunities through predictive analytics | 18% |
| Better customer experiences through faster responses and personalisation | 15% |
| Analysing demographic information | 13% |
| Improved pricing accuracy through deeper data insights | 12% |
Shifting Priorities: How is AI in Underwriting Evolving?
The top three priorities remain unchanged between 2024 and 2025: unstructured data analysis, task automation, and speed-to-market continue to lead. However, each saw a notable drop in percentage points, indicating more measured optimism or a levelling off following early adoption.
Notable declines
- Risk triage fell significantly from 62% to 38%.
- Managing complex queries dropped from 60% to 27%.
These shifts may reflect re-prioritisation as firms refine AI use cases or encounter implementation challenges.
New areas of focus in 2025
- Emerging and hidden risk detection: 28%
- Real-time underwriting: 20%
These gains suggest growing interest in proactive risk identification and responsiveness.
Meanwhile, portfolio insights, team capacity, and data enrichment have moved into the mid-tier of priorities.
A reduced emphasis on customer experience and pricing accuracy implies that AI in underwriting remains primarily operationally focused on backend efficiency rather than client-facing innovation.
Question 14: What are the best use cases for using AI within underwriting? (Top 3 were selected)
Answer:
| Use Case | % of Respondents |
|---|---|
| Process automation | 64% |
| Insights and analytics | 46% |
| Risk assessment | 42% |
| Data enrichment | 36% |
| Underwriter decision support and augmentation | 34% |
| Applying underwriting guidelines | 32% |
| Submissions | 27% |
| More accurate decision-making | 22% |
2025 Insights
Top use cases remain consistent, with a shift in focus emerging for 2025.
Process automation and insights & analytics continue to lead as the primary AI use cases in underwriting.
What's new in 2025:
- Underwriter decision support and augmentation has emerged as a key focus (34%), highlighting a growing trend toward AI enhancing—rather than replacing—human judgment.
Overall trend:
- Slight declines in percentage selections across most categories suggest a more measured, realistic view of AI's capabilities as adoption matures.
Question 15: What are the biggest challenges or pain points in implementing or adopting AI in underwriting workflows? (Up to five were selected)
Answer:
| Challenge / Pain Point | % of Recipients |
|---|---|
| Data concerns (quality or availability) | 55% |
| Regulatory/legal concerns | 52% |
| Integration with legacy systems | 39% |
| Cost of implementation | 37% |
| Balancing AI automation with the human element | 33% |
| Cybersecurity and privacy risks | 32% |
| Talent/skills gap (lack of internal expertise or skills) | 27% |
| Ethics concerns | 25% |
| Buy-in from the workforce | 24% |
| Finding the right vendor | 24% |
| Finding the best use case | 20% |
| Lack of proven ROI or use case success stories | 17% |
| Buy-in from the board/senior management | 13% |
| Model monitoring and maintenance | 12% |
| Not knowing where to start | 10% |
| Scalability | 9% |
| Operational disruption/business continuity risk | 8% |
Have pain points changed since 2024?
While data quality and regulation remain the biggest hurdles, their intensity has eased.
In 2025:
- Integration with legacy systems and cybersecurity risks have become more pressing.
- Cost and ethics concerns have declined.
- Talent shortages and finding the right vendors or use cases are now more significant barriers.
Overall takeaway:
The market is maturing, moving beyond awareness and early experimentation toward solving real-world implementation challenges.
Question 16: Which line of business has most to gain from leveraging AI? (Up to five were selected)
Answer:
| Line of Business | % of Respondents |
|---|---|
| Motor | 60% |
| Property (Commercial, Wholesale, Corporate, Investors) | 57% |
| Cyber | 43% |
| Casualty (Primary, Excess, General Liability) | 34% |
| Healthcare including Medical Malpractice | 32% |
| Parametric | 24% |
| Professional Liability (E&O, PI) | 21% |
| Aviation & Space | 20% |
| Cargo & Marine | 18% |
| Real Estate | 16% |
| Crop/Agriculture | 13% |
| D&O Liability | 11% |
| Environmental Liability / Pollution Legal Liability | 11% |
| Construction & Engineering | 10% |
| Legal Expenses | 8% |
| Terrorism | 7% |
| Energy | 6% |
| Latent Defects | 6% |
| Warranty & Indemnity (including M&A) | 5% |
| Surety | 5% |
| Forestry | 3% |
| Specie and Fine Art | 2% |
Insights: 2024 vs. 2025
While the top five lines of business remain largely consistent, Motor and Commercial Property have gained momentum in 2025.
This suggests AI's efficiency and automation potential is being most effectively realized in high-volume, data-rich underwriting areas.
Meanwhile, slight declines in Cyber and Healthcare may indicate maturing expectations or increasing concerns around implementation complexity and regulatory requirements.
Question 17: I expect AI to be part of my daily routine/workflow in underwriting within:
Answer:
| Timeframe | & of Respondents |
|---|---|
| Within the next 12 months | 30% |
| Within 1-3 years | 48% |
| 3+ years | 15% |
| 5+ years | 7% |
Question 18: How do you view the relationship between AI and human underwriters?
Answer:
| Response | % of Respondents |
|---|---|
| AI will enhance human decision-making and reduce workload, allowing underwriters to focus on complex cases | 55% |
| AI is a tool, but human underwriters are essential for all decisions | 30% |
| AI will replace many tasks currently done by humans | 12% |
| Unsure / Too early to tell | 4% |
Question 19: How is your organisation currently approaching insurance coverage for AI-related risks?
Answer:
| Response | Percentage |
|---|---|
| Not applicable / Don't know | 32% |
| Adapting existing policies (e.g., cyber, D&O, professional liability) to address AI risks | 20% |
| Believe current lines can cover AI risk, but are clarifying terms (silent coverage vs. exclusions) | 17% |
| Not yet addressing AI risk in our underwriting or risk transfer strategy | 17% |
| Considering or developing dedicated AI insurance products | 14% |
AI isn't replacing underwriting; it's changing how it's done. Survey results show that AI works best as a tool that helps underwriters spend more time on what technology can't execute. Find out how to deploy and govern agents across your underwriting safely and efficiently through our Agentic Framework.
Underwriting Resources
Underwriting Maturity Framework: Moving from a process-driven to a data-driven operating model
Grow your delegated underwriting
