Autonomous Finance in InsuranceWhy Data Quality Still Limits AI

Insurance’s autonomy paradox

The insurance sector is typically characterized as cautious when it comes to new technologies. Heavy regulation, long product lifecycles, and complex risk models have encouraged a measured approach to innovation, particularly within finance. But this caution hasn’t prevented widespread AI adoption.

According to the Global Autonomous Finance Benchmark 2025, a survey of 1,600 CFOs and finance professionals run in partnership between Aptitude, Microsoft and HSO, the insurance sector has jumped three places in AI usage rankings and now sits third overall. 97% of insurers report that AI is being used within their financial operations, indicating that adoption is no longer experimental.

On the surface, this should place insurance in a strong position to progress toward autonomous finance. However, the same benchmark data points to a more fragile reality. While AI capability has expanded, the conditions required to support autonomous decision-making remain disjointed. Progress is real, but not yet stable.

Autonomous finance refers to a finance operating model where financial data is captured, validated, and governed continuously as transactions occur. Reporting, controls, and decision-making operate in near real time with clear accountability and minimal manual intervention. 

In insurance, where financial outcomes must be defensible to regulators, auditors, and boards, applying autonomy without confidence in the underlying data increases risk rather than reducing it.

As Alex Curran, CEO of Aptitude Software, notes:

“Data is the key. I would encourage organisations to make sure that they have a finance data platform that holds vast quantities of information processed in a way that supports more sophisticated AI scenarios.”

What is autonomous finance in insurance?

Autonomous finance in insurance refers to a finance operating model where financial data is captured, validated, governed, and made decision-ready continuously. Reporting, controls, and forecasting operate in near real time, with minimal manual reconciliation and clear regulatory traceability.

AI adoption is widespread, but autonomy lags

Benchmark findings confirm that AI usage in insurance finance is now the norm rather than the exception. With 97% of insurers reporting active AI usage across functions such as reporting, forecasting, and analytics, the shift from experimentation to deployment is clearly underway.

However, widespread use has not translated into consistent operating maturity. AI capabilities are often introduced in parallel and attached to specific processes or teams. It is rare they are structurally integrated into a single finance operating model. As a result, maturity varies across activities, even within the same organisation.

Reporting may benefit from automated analysis, while forecasting still relies on manually adjusted data. Controls often remain dependent on reconciliation cycles that occur after the fact, placing trust in results outside AI-enabled workflows.

The result is a patchwork of improvements rather than a step change in how finance operates. Autonomy depends on consistency, with a shared foundation of trusted data across reporting, forecasting, and controls. Where this is lacking, AI remains an enhancement rather than an enabler of unbroken decision-making. In a regulated insurance environment, AI built on inconsistent data does not accelerate decision-making. It increases regulatory exposure and weakens confidence in financial reporting.

Insurance at a glance 

Insurance finance benchmark highlights:

• 97% actively using AI

• 52% say data quality limits decisions

• 25% identify as fully autonomous

• 43% cite transformation as a growth challenge

• 86% do not expect AI to reduce headcount

Why data quality is the real constraint on autonomy

Data quality has been a longstanding concern in insurance finance, and its impact is becoming more acute. 52% of insurers say that poor data quality and reliability are limiting their ability to make strategic decisions, an increase of more than 20% compared with 2024 and broadly in line with the global average.

In insurance, unreliable financial data undermines confidence in outcomes that must withstand regulatory scrutiny. When balances cannot be traced cleanly back to source transactions, or when assumptions differ across systems, trust in AI-generated insight collapses.

Poor data quality also amplifies risk. Forecasts can be produced faster, but if there is little confidence in the inputs, finance leaders are forced to validate results manually. Automation accelerates output but does not reduce effort. In some cases, it increases it.

Autonomous finance relies on financial data that is consistent, governed, and validated continuously before analysis begins, rather than corrected through reconciliation after the fact. Without that foundation, AI outputs remain provisional, decisions are delayed or challenged, and autonomy stalls.

Transformation has improved maturity, but increased strain

The benchmark shows genuine progress in insurance finance maturity. Firms operating with traditional finance models, characterized by siloed systems and manual processes, have fallen from 28% in 2024 to 13% in 2025. 

Over the same period, those identifying as fully autonomous, defined by access to real-time financial data with minimal human oversight, increased from 11% to 25%.

Despite this progress, 43% of insurers still identify digital transformation as a major growth challenge. Rather than a lack of activity, this reflects the strain created by overlapping initiatives.

It is common for multiple transformation programs to run concurrently, each introducing new platforms, processes, and data flows. While individual initiatives may improve efficiency in isolation, the combined effect is increased complexity. Ownership becomes blurred, dependencies multiply, and confidence in the numbers weakens even as work continues.

Here, transformation becomes a source of tension rather than momentum, particularly when data governance and ownership fail to evolve alongside new capabilities.

Why on AI adoption remains elusive in insurance finance

Despite high AI adoption, insurance firms continue to find measurable returns difficult to isolate. Major finance and technology programs typically take several years to deliver full value, especially in regulated environments.

In the interim, benefits such as time savings, improved accuracy, or reduced manual effort are absorbed into day-to-day operations rather than captured as discrete returns. The value of AI is real, but difficult to separate from ongoing activity when it is applied to data that still requires manual validation and adjustment.

This difficulty in separation helps explain why ROI lags adoption. Until finance teams can shorten the path from insight to action and rely on outputs without extensive validation, returns will remain thin.

What insurance CFOs want AI to change

For insurance industry CFOs, the benchmark shows that expectations around AI are pragmatic. 86% do not expect AI to deliver headcount reductions. Instead, they prioritize confidence, control, and decision quality.

For example, finance leaders want to reduce the effort consumed by reconciliation, exception handling, and results validation – activities which consume time without improving insight.

Talent pressures compound the issue. 49% of insurers say that talent acquisition and retention are challenging financial stability, a figure 9% higher than the global average. In this context, AI is expected to support judgment by making data more reliable and accessible, not by replacing expertise.

From AI-enabled finance to autonomous finance

Automation and autonomy are often conflated, but they place different demands on the finance function.

Automation improves efficiency by handling individual tasks faster or with fewer errors and can operate on fragmented data, provided outputs are checked first. Autonomous finance enables decisions to be made continuously, based on consistent data, and with minimal manual intervention.

Automation can tolerate inconsistency.
Autonomy cannot.

Where reconciliation remains a prerequisite for trust, AI outputs remain advisory and cannot safely support autonomous decision-making. Without continuous close, and without retaining financial data at a granular, event-level, autonomy cannot be sustained in a regulated insurance environment. 

True autonomy emerges only when finance teams rely on the same trusted data across reporting, forecasting, and controls.

At a glance – Insurance sector autonomous finance profile

• AI strategy ownership: Finance and IT

• Average time to value: 3–4 years

• Period-end close: 1–5 days

• Data processing frequency: Weekly

• Self-service reporting: Majority

Removing the structural barriers to autonomy

The main barriers to autonomous finance in insurance are operational rather than technical. Finance teams often remain dependent on upstream systems to prepare, aggregate, and adjust data before analysis can begin.

As well as extending delivery timelines, this dependency limits how quickly AI initiatives progress, and restricts finance’s ability to respond to change.

A finance-controlled approach to data reduces this friction. When finance owns how data is structured, governed, and prepared, AI can be applied consistently across use cases without waiting for upstream modifications. 

Over time, data trust improves, ROI accelerates, and AI begins supporting decisions in real time.

Fynapse solves the data problem

The barriers to autonomous finance stem from how financial data is prepared, governed, and made available to finance teams. If data quality and governance are the constraints, the solution must sit at the data layer, not at the AI layer.

Fynapse addresses this challenge by providing a finance-first data management platform that removes silos, structures financial data for AI use, and supports real-time insight without compromising control. Rather than relying on upstream system customisations, it gives finance direct ownership of how data is captured, validated, and governed.

Fynapse can be implemented in weeks rather than months or years, allowing insurance firms to shorten the gap between AI ambition and operational readiness. By removing data preparation bottlenecks, finance teams are able to apply AI consistently across reporting, forecasting, and controls, rather than in isolated pockets.

As Alex Curran, CEO of Aptitude Software, explains:
“It’s not about automating what already exists. It’s about reshaping how finance operates in a much more agile and insight-led way.”

Insurance organisations that have adopted Fynapse report faster implementation and clearer returns from their AI initiatives, including:

• Three times faster AI adoption by eliminating data preparation bottlenecks

• More than 50% reduction in manual reconciliation effort

• Over 4,000 FTE days saved through automation

By putting finance in control of its data, Fynapse creates the operating conditions required for autonomous finance to scale safely in a regulated insurance environment.

Autonomy requires confidence, not speed

The insurance sector has made substantial progress in adopting AI within finance, but that progress rests on foundations that are not yet strong enough to support autonomy at scale.

Confidence in data quality, governance, and ownership remains the defining constraint. Until these conditions are addressed together, AI will remain an overlay rather than a lever for sustained value.

Autonomous finance in insurance is achieved by creating operating conditions where accountability and trust are firmly established, financial data is governed continuously, and teams can act on insight immediately. It is not achieved by simply adding more AI.