How data and AI can transform insurance subscription.

For a long time, digital transformation in the insurance sector was primarily focused on sales channels, customer service, and client experience. Now, data and artificial intelligence are starting to advance over an even more strategic layer of the business: decision-making.

In underwriting, this movement is especially relevant. Evaluating risks requires gathering information from different sources, interpreting documents, identifying patterns, applying business rules, and often dealing with situations that do not fit perfectly into predefined models. It is precisely in this combination of data volume and decision-making complexity that AI finds space to generate value.

Results are already beginning to appear at different points in the insurance chain. In a project developed by Objective for a pension insurance company, for example, the use of artificial intelligence quadrupled predictive accuracy and contributed to generating R$ 342 million in incremental revenue. This example shows how well-structured data, combined with intelligent models, can evolve from merely providing operational support to directly participating in generating results for the business.

In underwriting, the potential follows the same logic. Enhanced underwriting and predictive risk assessment are among the main use cases of generative AI in the insurance sector. More than automating activities, these technologies can expand insurers’ analytical capacity, reduce the time needed for certain assessments, and offer more information for risk decisions.

However, this does not mean replacing the underwriter’s knowledge with an algorithm. The most relevant transformation lies elsewhere: using data and AI to enhance human decision-making capacity.

So, what is underwriting in insurance?

Underwriting is a central step in the insurance operation. It is in this process that the insurer assesses the risk associated with a person, company, or asset to decide whether to accept certain coverage and under what conditions.

To do this, the underwriter analyzes different information—such as history, policyholder profile, risk characteristics, and specific data for each type of coverage—to support decisions regarding acceptance, coverage, limits, and pricing.

In practice, the quality of underwriting directly influences the balance between risk and return in the portfolio. The more relevant information available and the greater the capacity to analyze it, the more consistent the decision-making tends to be.

It is precisely at this point that data and artificial intelligence begin to change the dynamics of underwriting.

How is AI transforming insurance underwriting?

Artificial intelligence enhances insurers’ ability to transform large volumes of data into useful information for risk analysis. In underwriting, this means combining historical data, documents, policyholder information, and other relevant sources to support decisions with greater context and consistency.

Machine learning models can identify patterns and relationships that are difficult to capture through traditional analyses, while generative AI adds new possibilities, such as interpreting documents, synthesizing information, and assisting the underwriter in analyzing more complex cases.

In practice, technology allows parts of the more standardized analyses to be automated, while situations that require greater interpretation can be sent to specialists, accompanied by information and recommendations to support their decision.

But the model’s capability directly depends on the data foundation it supports. As Objective executives emphasize in discussions about AI and data, AI does not solve data quality problems. According to Leandro Saran, COO, a structured and governed data architecture is a prerequisite for AI applications to scale safely.

Therefore, in underwriting, the evolution lies not just in adding AI to the existing process, but in building a reliable data foundation capable of making risk analysis increasingly intelligent.

What are the benefits of automated underwriting for insurers?

Automating part of the underwriting process allows insurers to assess risks with greater agility and consistency, reducing manual activities and increasing processing capacity without expanding the structure proportionately.

Another gain is in standardization of analyses. With well-defined rules and models, the insurer reduces variations in similar decisions and can direct underwriters to cases that truly require specialized knowledge.

The combination of data and AI also contributes to more accurate risk analyses and decisions that are more suited to each client’s profile. In practice, this can represent greater operational efficiency, better risk selection, and increased scalability for the business.

What are the challenges of implementing AI in insurance underwriting?

The first challenge lies in the data. Fragmented information, inconsistent historical databases, or poor quality of records can compromise the models’ ability to assess risks reliably.

Another point is the integration with legacy systems. In many insurers, data and underwriting processes are distributed among different platforms, making the operation’s modernization more complex.

There are also issues related to governance, explainability, and biases. Since underwriting decisions can directly impact the acceptance and terms of insurance, it is essential to understand how models arrive at their recommendations, monitor their performance, and establish criteria for safety, privacy, and compliance.

In this context, finding a partner with experience in data, AI, and complex environments is also part of the challenge. Objective supports companies in structuring their AI governance, connecting technology to business needs and establishing mechanisms for risk management, data quality, security, compliance, traceability, and model monitoring.

More than creating controls, the goal is to establish a foundation that enables scaling AI safely and responsibly, without turning governance into a barrier to innovation.

How to balance automation and human judgment in underwriting?

The balance between automation and human decision-making begins with the design of the process. Not every step in underwriting requires the same level of analysis, and it is essential to define which decisions can be automated, which need validation, and which should remain under the underwriter’s responsibility.

One possible approach is to structure the workflow by levels of complexity and risk. Standardized cases that meet predefined criteria can proceed through automated processing. Situations presenting discrepancies, low model confidence, or falling outside established rules are forwarded for human analysis.

For this model to function, several elements need to be integrated into the process:

• business rules and decision thresholds, defining the limits of automation;
• scores and confidence levels, which help determine when a recommendation can advance automatically;
• exception management, directing outlier cases to the appropriate specialist;
• audit trails, recording the data used, model recommendations, and decisions made;
• continuous monitoring, comparing model performance with the actual results of the portfolio.

The underwriter thus acts primarily where their knowledge adds the most value: complex cases, exceptions, and decisions requiring context. At the same time, human decisions can feed back into the process, creating data to evaluate and continuously improve rules and models.

The goal is not to choose between automation or human judgment, but to design a process where each decision is directed to the appropriate level of automation, control, and specialization.

Where to start transforming underwriting with data and AI?

Transforming underwriting does not need to begin with the automation of the entire process. The first step is to understand the current scenario: how decisions are made, which steps require the most manual effort, where the main bottlenecks are, and what data is available to support evolution.

From this diagnosis, it is possible to prioritize use cases with potential to generate value, assess data and architecture maturity, define outcome indicators, and test solutions in a controlled manner before scaling their use.

This pathway also allows for establishing governance from the start and gradually evolving automation levels as models demonstrate performance, reliability, and adherence to business rules.

More importantly, however, is understanding that there is no one-size-fits-all formula for transforming underwriting with AI. Each insurer has different processes, systems, data, products, rules, and maturity levels—and technology needs to respond to this reality.

At Objective, each journey is built from the context and specific challenges of the company, combining strategy, data, artificial intelligence, architecture, and development to create customized solutions capable of evolving with the business.

Want to understand where data and AI can generate more value for your operation? Talk to the specialists at Objective. We can analyze your scenario, identify opportunities, and build a solution tailored to your business needs.

Frequently asked questions on the topic