A form has never been the best way to understand someone.

From minimal input to rich context: multimodal input is changing how financial institutions make decisions.

Turner provides its clients with timely, fresh, and relevant insights into its markets and the Strategy Execution challenges they face. In this blog series, Frits Korthals Altes, Director of Digital Transformation, AI & Innovation at Turner, explains why exceptions are no longer more expensive than standard cases, how standardization is shifting toward data, infrastructure, and AI capabilities, and what design choices are currently on the table.

This blog explores the shift from minimized datasets to context-rich information processing, the role of multimodal AI in this shift, and the implications for decision-making processes, customer experience, and fraud management at insurance companies and pension funds.

“People are much better able to explain their situation and the challenges they face in a conversation than on a form.”

The form was a compromise, not the best solution

For decades, financial institutions have organized their data collection around a single principle: ask as little as possible, in as structured a way as possible. Not because that was the best way to understand a customer or a situation. But because data collection was expensive, data entry was error-prone, and systems could only function with standardized and structured data fields. The result is an industry full of forms. Dozens of fields for a mortgage application. Long questionnaires for disability claims. Claim forms that reduce reality to a few checkboxes and a description of no more than 500 characters. Information that didn’t fit into that structure was omitted, even if it was relevant. That came at a cost. A customer who describes a complex claim in three lines of text provides less information than the same customer who speaks for three minutes. An insured person who selects their employment status from a drop-down list communicates less than someone who explains what’s really going on. The limitation wasn’t with the customer. It was in the system. That limitation has now been removed.

Multimodal AI processes what people naturally share

What is changing structurally is not that AI automatically fills out forms. The mechanism is more fundamental: AI can instantly process any form of input and convert it into usable information. Text, speech, images, video—all modalities become equally valuable and combinable. Frits Korthals Altes describes the essence of this shift as follows: “When you talk to people, they’re much better able to explain their situation and the challenges they face than they are on a form.” That insight has direct implications for the process. The most informative input isn’t the most structured input. It’s the input that most closely aligns with how people naturally communicate. A disability insurance provider in the Netherlands has already put this principle into practice. Instead of filling out a detailed questionnaire, the insured person records a spoken statement. They describe their situation, what they’re still able to do, and what daily obstacles they face. This spoken explanation provides richer, more nuanced information than any form could, and it lowers the barrier for the customer. The system adapts to the person, not the other way around. The same principle applies to claims processing. A short video of the damage provides information on the extent, cause, context, and materials that an on-site inspector would normally have to gather, but which is now immediately available. The combination of video, spoken commentary, and metadata provides a more complete picture of the situation than any structured form can offer.

Three implications for processes that now need to change

Anyone who takes multimodal input seriously as a design principle reaps three benefits:

First: the intake becomes a conversation, not a form to fill out. Customers provide information through the method that feels most natural to them. This lowers the barrier to entry, increases the completeness of the information, and reduces reliance on customer motivation to work through long forms.

Second: decision quality improves through more context, not through more fields. A decision to approve a disability claim based on a verbal explanation, supplementary medical information, and behavioral data is more accurate than the same decision based on a completed questionnaire. More context means less residual risk and fewer incorrect denials.

Third, fraud detection benefits directly from contextual richness. A claim report that contains inconsistencies between the spoken explanation, the video recording, and the written description contains signals that no structured form can capture. Multimodal AI automatically recognizes these inconsistencies and flags them for further investigation, without the customer even noticing.

A disability insurance provider showed that there are other ways to do things

A real-world example illustrates what is possible once the form is no longer the starting point. A Dutch disability insurance provider replaced the traditional questionnaire used during claim intake with a voice recording model: the insured person explains in their own words what their situation is, what they are still able to do, and what obstacles they face in their daily life. The benefits were manifold. The qualitative information obtained was richer and more nuanced than the questionnaire had ever provided. The barrier to participation for the insured person visibly decreased; filling out a long questionnaire requires more motivation than a three-minute voice recording. And the assessor had more context available to make an accurate decision. What this example also shows is that technology is not the bottleneck. The willingness to redesign the intake process around what the customer naturally does is the bottleneck. That is a strategic choice, not a technical one.

Three Key Considerations for Directors

Multimodal intake raises issues that require consideration before decisions are made.

The first consideration is consent management. Audio and video recordings of clients are special categories of personal data under the GDPR. An institution that implements multimodal intake without an explicit consent framework and data retention policy builds a privacy risk into the core of its claims process.

The second point to consider is model interpretation for subjective modalities. AI that interprets speech, intonation, or video relies on statistical patterns. These patterns may unintentionally and systematically deviate for certain accents, dialects, or cultural communication styles. An underwriting decision that is partly based on multimodal interpretation must be explainable, even if that interpretation yields unequal results for different customer groups.

The third point to consider is the illusion of completeness. More detailed input does not automatically lead to better decisions if the model or the evaluator is not designed to make use of that detail. An organization that implements a multimodal intake process but does not adjust its decision-making logic ends up collecting more data without achieving better outcomes, while simultaneously making the customer experience more complex. The question is not whether the form will disappear, but when.

Three Insights

A limited dataset was a technical necessity, not a substantive principle. The limitation that influenced the process architecture for decades has been lifted. Institutions that fail to translate this into a redesign are voluntarily creating an information backlog.
Rich, multimodal input leads to better decisions, less friction, and more effective fraud detection all at once. These are not separate aspects; it is a single design principle with three benefits.
Anyone who retains the form as the primary intake tool is making a strategic decision—even if that decision isn’t explicitly stated. An institution that, three years from now, is still working with standardized questionnaires while competitors and new entrants are processing rich context isn’t lagging behind in technology but in decision quality.

Three questions for your organization:

Do you know which of your core processes suffer from a structural lack of information from the current intake mechanism, and who in your organization is responsible for redesigning it?
In your claims handling or underwriting process, is there currently someone responsible for the overall quality of decision-making, including the completeness of the data?
What consent and privacy architecture do you have in place if multimodal intake becomes part of your core processes over the next two years?

Most insurers and pension funds are aware that their current intake mechanisms are missing information. Fewer know what design choices to make to change that without introducing the associated compliance risks.

AI Series

This blog is part of the series “Successful AI Implementation in Financial Services”

Eleven blogs that show you how to truly harvest the benefits of AI. Here are three suggestions for further reading.

Back to the overview

Show the entire series

AI isn’t a tool you simply roll out. It’s a mindset you need to cultivate.

A form has never been the best way to understand someone. AI enables multimodal input.

Want to learn more about Turner Financial services?

Banks, insurers, and pension funds are indispensable. How can they continue to deliver value in an uncertain future? Turner Financial Services helps financial institutions improve, innovate, and accelerate Strategy Execution.

Would you like to improve, innovate, and transform faster?

Turner is specialized in strategy execution. We make it our business to turn an organisations strategy into reality. For the seventh consecutive year, Turner has been recognized by clients as the best strategic consulting firm in the Netherlands. More than 4,000 executives and professionals participated in the MT1000 survey, in which Turner received the highest ratings for strategic consulting, client focus, excellent execution, and product leadership.

Read more about the results other clients achieved with the support of Turner Strategy = Execution. Ready to talk? Schedule an appointment. Contact us.