InsTech - insurance & innovation with Matthew Grant & Robin Merttens

In just twelve months, the conversation around Agentic AI in insurance has changed dramatically. What began as curiosity about autonomous AI agents has evolved into a much more practical discussion about implementation, governance, economics and competitive advantage. In this special solo episode, InsTech's Zoja Wojcik reflects on the developments that have shaped the market since InsTech's first Agentic AI event in November 2025. Drawing on conversations with insurers, brokers, MGAs, technology providers and industry leaders, she explores how the industry has moved beyond experimentation and towards a more challenging question: where does the commercial value actually come from? Along the way, you'll hear insights from Simon Torrance, Erdal Atakan, Gina Gill, Elena Maran, Max Richter and Ian Thompson, alongside examples of how organisations including CFC, McGill & Partners, AIG, Duck Creek and hyperexponential are bringing Agentic AI into real insurance operations. Whether you're still trying to understand what Agentic AI means for insurance or already evaluating deployment opportunities, this episode offers a practical snapshot of where the market stands today and the questions leaders should be asking next. Want to continue the conversation? Join us in London on July 7 for 'The age of Agentic AI: from strategy to commercial value'. In this episode: 00:00 - What is Agentic AI and why has it become one of insurance's most discussed technologies? 03:15 - Looking back at the industry's first major Agentic AI event in November 2025 05:45 - Simon Torrance on why Agentic AI should be viewed as a new workforce, not simply another software tool 06:20 - Early deployment examples from across the insurance market: CFC's Lane Assist McGill & Partners and Salesforce Agentforce AIG's AI-driven underwriting initiatives Federato's agentic underwriting platform hyperexponential and Banyan Risk Duck Creek's insurance-native Agentic AI platform 08:15 - Why moving from pilot projects to production remains difficult 10:00 - The defining question of 2026: proving commercial value and ROI 12:15 - Intelligence Capital, competitive advantage and why buying AI tools may only create parity 13:30 - Orchestration, governance and maintaining trust in agentic systems 15:00 - Workforce transformation and practical lessons for insurance leaders 16:00 - What questions should insurance organisations be asking next? Key takeaways: The industry conversation has shifted from experimentation towards implementation and measurable business outcomes. Many of the biggest barriers to adoption are organisational rather than technical. Boards increasingly expect clear economic justification for AI investment. Competitive advantage may come less from AI models themselves and more from institutional knowledge and decision-making expertise. Governance frameworks must evolve alongside increasingly autonomous systems. Organisations that focus on specific business problems are more likely to succeed than those pursuing AI for its own sake. Featured contributors: Simon Torrance, AI Risk Erdal Atakan, Inigo Gina Gill, Apollo Elena Maran, Alethesis AI Max Richter, Mea platform Ian Thompson, IMT Advisory Further reading: For listeners looking to explore the themes discussed in this episode: Agentic AI & insurance Podcast episode: Where is the industry today? – a view from the C-suite (A rare C-suite perspective on Agentic AI: what it is, how it’s being deployed and why senior leaders are walking a tightrope between bold innovation and operational risk.) CFC launches Lane Assist, a live agentic underwriting pilot McGill & Partners becomes first London Market broker to deploy Agentic AI McGill + AIG collaboration using AI-driven underwriting Duck Creek launches insurance-native Agentic AI Platform Federato RiskOps and Agentic underwriting platform MGA Banyan Risk deploys hx’s full agent

In this episode, Matthew Grant is joined by Richard Louden, Partner (Indirect Tax Financial Services) at KPMG, and Candy Staples, Director (Innovation Reliefs and Incentives), to explore a topic that many insurance and InsurTech businesses underestimate until it becomes expensive: tax. While tax is often viewed as a back-office concern for finance teams, it can have a significant strategic impact on how insurance businesses operate, scale and ultimately exit. From the complexities of VAT and Insurance Premium Tax (IPT) to the opportunities created by R&D tax incentives and the Patent Box regime, the conversation highlights both the risks of getting tax wrong and the upside of approaching it proactively. Richard brings more than two decades of experience advising insurers and intermediaries on indirect tax. He explains why VAT behaves differently in insurance compared with most industries, and why misunderstandings around exemptions, commissions and international services regularly create costly problems for growing businesses. Candy focuses on the more positive side of the equation: how innovation incentives can help companies recover the cost of developing new technology. For InsurTech firms investing heavily in product development, these incentives can represent a meaningful source of funding and cash flow if captured correctly. At the heart of the discussion is a simple message: tax is not just about compliance. Managed properly, it can influence profitability, operational efficiency and investment decisions across the insurance value chain. In this conversation, Richard and Candy share: Why VAT behaves differently in insurance and why exempt supplies can quietly increase operating costs The common misconception that commission structures automatically determine VAT treatment How the reverse charge mechanism on overseas services often creates unexpected liabilities Why start-ups have a strategic advantage when designing tax processes from day one How R&D tax credits can return meaningful cash to companies investing in innovation Why capturing technical challenges and development work early is critical for successful claims How the Patent Box regime can reduce corporation tax on profits linked to patented technology Why tax incentives should be considered alongside broader decisions about where companies locate teams, IP and development hubs KPMG are also hosting post-ITI drinks in London with Insurtech UK to navigate the headwinds of today's economic and regulatory challenges facing insurers and insurtechs alike over cocktails, food and conversation. Click here to register your interest: https://insurtechuk.org/events/0319-one-last-stop-from-headwinds-to-happy-hour/ Additionally KPMG Actuarial have released a white paper on Smarter Pricing, Smarter Insurance. How integrated data, AI and governance transform underwriting and growth. Download to read here: https://m.marketing.kpmg.uk/webApp/Smarter_pricing_Smarter_insurance_whitepaper If you like what you’re hearing, please leave us a review on whichever platform you use or contact Matthew Grant on LinkedIn. Sign up to the InsTech newsletter for a fresh view on the world every Wednesday morning.

In this episode, Matthew Grant sits down with Nicola Turner and Alex Ley, co-founders of Scrub AI, to explore one of the most pressing strategic questions facing insurers today: Build vs Buy in the age of generative AI. Five years into building an AI-driven data cleansing platform for carriers and brokers, Nicola and Alex have seen the market shift from scepticism to urgency. Boardrooms are now asking how AI is being embedded into underwriting workflows, and whether those capabilities should be developed internally or sourced from specialists. Drawing on their experience building deterministic AI models for exposure data and catastrophe modelling, they offer a grounded perspective on what works, what breaks and where the real risks sit. At the heart of the discussion is a simple truth: getting to 80% is easy. Getting the final 20% right is where strategy, domain expertise and long-term thinking matter most. In this conversation, Nicola and Alex share: Why Build vs Buy has intensified as generative AI moves from experimentation to executive priority How investor pressure and board-level scrutiny are shaping AI strategy inside large carriers Why generative AI can accelerate development but does not remove the complexity of insurance data The danger of plausible but wrong outputs in exposure management and catastrophe modelling Why deterministic AI still plays a critical role in delivering consistent, renewal-ready data How inconsistent data cleaning can distort underwriting decisions and renewal pricing The hidden cost of technical debt when insurers attempt to build internally Why maintaining and iterating ai tools is often harder than building the first version If you like what you’re hearing, please leave us a review on whichever platform you use or contact Matthew Grant on LinkedIn. Sign up to the InsTech newsletter for a fresh view on the world every Wednesday morning.

In this episode, we bring you a live panel from InsTech’s May event at CodeNode, exploring how automation is reshaping claims in the Lloyd’s and London market — and why the belief that specialty is too complex to automate no longer stands. Moderated by Matthew Grant, CEO of InsTech, the panel features Simon White, Chief Claims Officer at Apollo, Aidan O’Neill, Founder and CEO of DOCOsoft, and Zoe Woods, Claims Improvement Manager at Lloyd’s. Specialty claims have long been viewed as too bespoke, too nuanced and too reliant on human judgement for automation to play a meaningful role. But as underwriting becomes algorithmic and distribution turns digital, claims can no longer lag behind. This conversation moves beyond theory to evidence. Automation is already embedded in live workflows across the market. The firms adopting early are seeing measurable operational gains. In this conversation, they share: Why the myth that specialty claims cannot be automated is finally breaking down How Apollo processed more than 23,000 claims through automated checks, cutting handling times to under a working day What happens when you ask claims handlers to map every task they repeat on each file Why automation should augment decision-making rather than create black boxes How structured data and integrated dashboards unlock meaningful AI use cases What Lloyd’s is doing to balance innovation with oversight in a syndicated market Why modular, plug-and-play services are replacing large-scale transformation programmes What specialty can learn from automation in motor and property lines Why starting small with repeatable processes creates fast, tangible wins How claims is shifting from cost centre to strategic differentiator If you like what you’re hearing, please leave us a review on whichever platform you use or contact Matthew Grant on LinkedIn. Sign up to the InsTech newsletter for a fresh view on the world every Wednesday morning.

What happens when AI meets the backbone of the insurance industry - policy administration systems? In this episode, Liselotte Munk, CEO of Fadata, joins Robin Merttens to unpack how artificial intelligence is reshaping the software layer of insurance. With candid insights into Fadata’s AI strategy, Liselotte reveals how the company is using AI to accelerate software development and reduce implementation costs while improving quality. She tackles the big question: will AI make policy admin systems obsolete? Her answer offers a pragmatic view on cost, complexity, compliance and collaboration. In this conversation, Liselotte shares: How AI is already streamlining configuration, documentation and testing in core systems Why the true opportunity lies in faster implementations and reduced transformation costs How the role of developers is shifting, and what this means for insurance talent Why insurers should invest in AI to enhance - not replace - their core platforms What the smartest insurers are doing now to future-proof operations in an AI-first world If you like what you’re hearing, please leave us a review on whichever platform you use or contact Robin Merttens on LinkedIn. Sign up to the InsTech newsletter for a fresh view on the world every Wednesday morning.

This week we bring you an episode of Bootstrap Confidential featuring InsTech’s very own CEO, Matthew Grant, who joined Charles Green in the latter half of 2025 to reflect on the eight-year journey of building InsTech from the ground up without outside funding, and with an intentional focus on sustainable growth. Matthew’s route to growing InsTech wasn’t typical. With a background in risk, analytics and around 400 podcast episodes as a host, he brought a deep understanding of the insurance sector and what it takes to build a commercially viable, insight-led business. The result? A thriving community of over 30,000, a high-margin membership model and a successful exit achieved through discipline, focus and clear-eyed decisions. In this conversation, Matthew shares: Why he sees bootstrapping as risk management, not risk taking The importance of paying yourself from day one and how that shaped InsTech’s trajectory Lessons from testing (and killing) products that didn’t deliver Why hiring curious, early-career talent paid off What most founders get wrong about option schemes and equity How to handle financial stress without losing your team or your sanity Why co-founders matter and why investors aren’t a substitute The hard truth about building the business your customers want, not just the one you’d love to run If you like what you’re hearing, please leave us a review on whichever platform you use or contact Matthew Grant on LinkedIn. Sign up to the InsTech newsletter for a fresh view on the world every Wednesday morning.

How can AI weather models improve the accuracy and scale of catastrophe modelling? Matthew Grant is joined by David Wood, Managing Director at JBA Risk Management, and Jochen Papenbrock, Head of Financial Technology (EMEA) at NVIDIA, to explore how accelerated computing is unlocking new ways to simulate and manage flood risk. JBA has long been a pioneer in flood modelling, while NVIDIA’s GPU technology has helped drive the recent breakthroughs in AI and generative modelling. Together, they discuss how high-resolution simulations, new ensemble methods and open-source tools are pushing the limits of what’s possible in climate and catastrophe analytics. Key Talking Points: The early bet – how JBA’s adoption of GPU computing over a decade ago made national-scale flood mapping possible From gaming to GenAI – how NVIDIA's evolution from graphics to AI led to the development of physics-informed weather models Ensemble power – why running 1,000+ simulations helps capture more extremes than the historic record ever could Event sets reimagined – how AI models are enabling richer, more diverse flood scenarios for Europe and beyond Real-time relevance – the potential to use AI models to simulate how a flood might unfold, as it’s happening Making AI usable – how Earth-2 Studio and open-source frameworks are opening up generative models to catastrophe modellers Proving value – how NVIDIA and JBA worked together to quantify the benefits of faster, more flexible modelling approaches Looking ahead – why cross-sector collaboration will be essential to turn acceleration into real-world impact If you like what you’re hearing, please leave us a review on whichever platform you use or contact Matthew Grant on LinkedIn. Sign up to the InsTech newsletter for a fresh view on the world every Wednesday morning.

In this episode, Claire Souch is joined by Tom Philp, CEO of Maximum Information; James Lay, AVP of Product Management at Verisk; and Stephen Martin, Head of Catastrophe Modelling at Westfield Specialty, for a timely discussion on the future of catastrophe model evaluation, and why it's no longer enough to simply trust what’s in the black box. As new specialist model vendors emerge and market expectations evolve, the panel unpacks a growing demand for transparency, interoperability and smarter ways to adopt models that fit real-world portfolios. At the heart of the conversation is a shared belief: the industry doesn’t just need more models, it needs better ways to evaluate and use them. In this conversation, they explore: Why traditional model validation no longer meets the needs of modern risk teams The shift from 'black box' outputs to meaningful model evaluation that supports business decisions How tools from Maximum Information and Verisk’s Model Exchange reduce the burden on small or lean teams The role of Oasis as a framework for opening up access across multiple model vendors Why standardisation and open data formats are essential for meaningful interoperability The growing role of niche vendors in reshaping perceptions of model transparency How automation is changing the regulatory and investor reporting game Why this is more than a tech upgrade—it's a cultural reset in catastrophe modelling Sign up to the InsTech newsletter for a fresh view on the world every Wednesday morning.