Many organisations are experimenting with tools that can ingest information, uncover patterns and support more informed portfolio decisions.
Nevertheless, much of the market still relies on repeated manual processes, with each firm cleansing data separately. This leads to inconsistent inputs and underwriting decisions, and valuable insight often remains trapped in isolated systems. A shift is beginning to occur, though progress varies across the market.
To better understand how this shift is playing out in practice, we sat down with Bernadette Tredger, Head of Portfolio Management at Apollo’s Smart Follow initiative who has over 20 years of experience in the London market, as part of our IT Insights InsurTalk series.
Bernadette’s experience spans actuarial science, underwriting strategy and data-led portfolio management, giving her a broad perspective on the role of AI in today’s market. As Bernadette noted in our recent chat,
AI can execute much of the underwriting workload… but we emphasise strong links with producers of business
The balance between technical capability and human oversight is becoming one of the defining issues for insurers as they consider the next stage of their development.
The conversation reflects a market that is searching for ways to work faster, make better use of data and align technology with the judgement that sits at the heart of underwriting, particularly in Smart Follow capacity.
Why Smart Follow capacity is ripe for transformation
Smart Follow capacity has always been an essential part of the London market. It refers to the portion of capacity provided by insurers that do not lead a placement but choose to follow the terms agreed by the lead underwriter.
These followers accept the rate, wording and conditions already negotiated by the lead and participate on that basis, allowing the market to syndicate large risks efficiently, although it also means many organisations work from the same underlying information. That is where the strain begins to show.
Revolutionise your claims operations with futureClaims™
futureClaims™ is an advanced platform designed to meet the demanding requirements of complex commercial and specialty claims, including the London Market.
A single risk often passes through multiple firms, each one repeating much of the same work. Data is cleansed and restructured again and again, models are rebuilt from scratch and exposure checks are carried out in isolation. As Bernadette highlighted, “everyone cleans the data, everyone models”, which captures the level of duplication that still shapes many underwriting decisions.
This repetition is not simply inefficient, it fragments insight and slows decisions. It also means different firms often reach different conclusions from the same information. With specialist talent already stretched across financial services, the market cannot sustain a model where the same technical work is rebuilt in every syndicate.
The growing use of AI-driven submission processing and data enrichment tools offers a practical alternative. Instead of every firm repeating the same work, these tools centralise data capture and enrichment, reveal patterns often missed in manual review and flag potential exposure clashes earlier. When shared with trusted partners, the benefits scale: firms gain a consistent view of performance, brokers get faster answers and lead underwriters can defend and grow their books without being constrained by data volume.
Read more about underwriting on our blog:
This is where much of the current progression is gathering momentum. The shift towards shared enrichment and augmented insight has the potential to raise standards across the follow market while still allowing individual firms to apply their own judgement and appetite. It marks a move away from a model defined by repeated manual effort, and towards one where improved data flow and consistent inputs create space for more strategic, portfolio-led underwriting.
How AI augmentation is redefining underwriting in the insurance market
The early conversation around AI in insurance often centred on the idea of full automation, with machines replacing large parts of the underwriting process. That expectation has faded as the practical limits of automation have become clearer. The value now lies in augmentation, where technology strengthens the quality and speed of underwriting without removing the judgement that experienced practitioners bring to the table.
Bernadette explained this shift clearly in our discussion, stressing that “there will still be people, this won’t be computers running themselves”. AI can collect and restructure information far faster than a human team, and it can reveal links across submissions that might remain unnoticed through manual review. What it cannot do is apply context, weigh degrees of uncertainty or understand how broader market conditions influence a risk. Those decisions still depend on human expertise, and the most effective AI models are being developed to support that expertise rather than replace it.
This is particularly relevant in Smart Follow capacity, where success depends on the ability to respond to submissions quickly while maintaining a consistent view of appetite and exposure.
Augmented underwriting helps achieve this by creating a more reliable flow of information into the underwriting process. With a clearer and more structured set of inputs, underwriters can concentrate on the points that matter most - from the strength of the lead to the pattern of accumulation across a portfolio.
The progression now taking place in the market is shaped by this balance. Firms are moving away from the idea of automation as an end goal and towards a model where AI provides the analytical foundation and humans carry out the interpretation.
The result is a more transparent approach to underwriting, where assumptions and data inputs are visible and traceable, and a more resilient one, where standardised, AI-supported workflows reduce manual exposure points and keep underwriting decisions consistent even during high-volume or resource-limited periods.
The rise of portfolio-led underwriting in the AI-enabled insurance market
A key sign of progress in the follow market is the move from assessing individual risks to viewing decisions at portfolio level. Firms are now looking at how each placement affects exposure, class balance and rate movement. AI speeds this up by providing quick access to structured data and revealing trends that once took days of manual analysis.
Bernadette explained how this works in practice. Bound risks are ingested quickly so portfolios can be updated and measured against expectations, creating a feedback loop where post-bind insights inform new business. This reduces reliance on intuition by highlighting appetite drift, emerging accumulations and areas where capacity can be added without increasing volatility.
As Bernadette explained, the real question becomes whether you are “overweight in ‘blue ships’ versus ‘red ships’”, rather than whether a single risk looks attractive in isolation.
Streamlining the claims underwriting process with an MVP integrating disparate data sources into a single system
Our MVP will enhance data accessibility, improve user experience and operational efficiency for claims underwriters, enabling future AI-driven developments, including data synthesis and process automation.
The discipline behind this model relies on more than data. Exposure limits, appetite rules and clearly defined controls still shape underwriting choices and help prevent outsized positions from forming. Many firms also use a blend of vanilla and more specialist risks to soften the impact of volatility, particularly in classes where loss patterns are uneven. AI plays a role in surfacing these relationships, but the decisions that follow remain firmly in human hands.
Current market activity supports this move towards portfolio-led underwriting. Technology can enhance facilities by encoding rules, automating referrals and providing real-time dashboards which gives underwriters far greater visibility over performance. They can see how each new risk affects the whole book, adjust their stance when deviations appear and keep brokers informed without slowing the placement process.
The progression taking place here is less about automation and more about strengthening the connection between information, appetite and action. With better data and quicker insight, underwriters can manage portfolios with more confidence and consistency, which is increasingly important as competition intensifies across the follow market.
The barriers slowing AI adoption in the insurance and underwriting sector
For all the enthusiasm surrounding AI, genuine progress in the insurance market is still held back by several structural constraints.
Data scarcity and incomplete foundations
The first barrier is the scarcity of consistent, high quality data. Many firms start new initiatives without the depth of information needed to build reliable models or justify performance expectations. As Bernadette noted, success begins with being “open minded about what is out there” and knowing how to identify benchmarks and structured datasets that can support augmentation. Without this foundation, even advanced tools deliver limited value.
A shortage of technical and domain expertise
A second barrier is the limited availability of people who understand both insurance and modern technology. Data science and engineering skills are in short supply across financial services, and this gap becomes more visible when firms try to integrate AI into underwriting or portfolio management. The market cannot sustain a structure where each firm builds and maintains its own complex data processes; the cost is too high and the expertise too scarce to solve the same problems repeatedly.
Caution, governance concerns and market perception
The third hurdle is market perception. AI-enabled underwriting is still new, and many organisations are cautious about adopting models that alter established workflows. Concerns around governance, accuracy and transparency shape how quickly firms are willing to move, especially where decisions carry regulatory consequences. This contributes to uneven adoption across the London market, with some organisations advancing rapidly and others taking a more measured approach.
How the market is beginning to respond
The emerging progression reflects these barriers. Actuarial benchmarks combined with structured data allow firms to build evidence frameworks that show expected portfolio behaviour.
Shared enrichment cuts duplication and improves consistency, strengthening oversight and supporting clearer decision-making. Appetite-rule platforms help leads maintain control while benefiting from smoother information flow.
Collectively, these changes demonstrate how the market can overcome earlier constraints and adopt AI in a controlled, credible way without losing the accountability and judgement that define the London market.
The Future of AI-driven underwriting and follow capacity in the London Market
The direction of travel in the follow market is becoming clearer. Early discussions often imagined fully autonomous underwriting, but the credible path blends augmented insight with strong governance.
Technology delivers speed and consistency, while underwriters retain control and apply the judgement that gives those insights value.
Bernadette emphasised this long-term view:
There will still be people; this will not be computers running themselves.
That sentiment is widely shared. Investors, regulators and brokers expect clear visibility into how decisions are made – a demand that will grow as portfolios become more data-heavy and AI assumes a larger share of the workload. Firms that can document material decisions, show where human oversight sits and explain model behaviour will be better placed to maintain confidence.
Competitive advantage will move toward firms that refresh portfolios quickly and manage exposure with clarity. Clean data pipelines, fast ingestion and reliable enrichment form the base. Appetite-rule platforms and performance-tracking tools give underwriters the visibility needed to respond to change without being overwhelmed.
As this progression continues, the underwriter’s role will evolve. Less manual review creates more space for thinking about portfolio shape, accumulation and strategy. Decisions remain with people, but the information supporting them becomes faster, clearer and more consistent. For firms that combine AI, data and expertise effectively, this points to a more agile and resilient underwriting model.
Summary
Our conversation with Bernadette highlights a market that is progressing, although at different speeds. Smart Follow underwriting remains central to syndication in London, yet it still carries duplicated processes and inconsistent data handling. AI is helping address this by improving ingestion, enrichment and exposure visibility, reducing manual effort and improving the quality of inputs.
The shift from automation to augmentation sits at the heart of this change. AI improves the information underwriters work with, but judgement and oversight remain essential. As more firms adopt portfolio-led approaches, underwriters gain a clearer view of how each risk affects the wider book, enabling quicker and more controlled adjustments.
Progress is still shaped by barriers such as limited data, scarce technical talent and governance concerns. The firms moving fastest are those building strong foundations with structured datasets, benchmarking and shared enrichment, creating an environment where innovation and accountability align.
To explore these themes in more detail you can watch our full discussion with Bernadette as part of our IT Insights InsurTalk series.
With 25 years of experience in data harmonisation, digital engineering, consulting, AI and cloud solutions, Future Processing helps organisations modernise their technology landscape and deliver complex digital initiatives with confidence.
If you would like to discuss how our expertise can support your organisation’s transformation goals, contact us today and we will work with you to find the right solution for your needs.
Future Processing works with organisations across many sectors to develop practical AI and data capabilities that improve decision making and reduce operational complexity.
If you are exploring similar challenges, we would be happy to discuss how our experience can support your goals and help shape a clear path forward.
