Behavioural AI on the Rise: The Power of Listening Differently

Across Southeast Asia, artificial intelligence is evolving from predictive automation into a tool for deeper human understanding. Institutions are adopting behavioural AI not just to process data, but to interpret behaviour, including how people speak, engage, and express emotions. Voice analytics is emerging as a key enabler, analysing vocal elements like tone, pace, and rhythm to uncover insights into stress levels, intent, and emotional well-being.

At the recent Asian Conference of Criminal and Operations Psychology, (ACCOP 25), organized by the Singapore Ministry of Home Affairs (MHA), this transformation and real life examples of AI usages were on full display at the event.

Among the AI pilot programs that was selected for a demonstration and conversation with Ms. Sim Ann, Senior Minister of State (Ministry of Foreign Affairs and Ministry of Home Affairs), and Mr. Pang Kin Keong, Permanent Secretary (Home Affairs), was a project developed by the Singapore Prison Service in conjuncture with a Singapore-based AI company, Voicesense, showcasing how behavioural voice analytics can be applied to support rehabilitation.

As agencies across the region explore new applications of AI for human behaviour analysis, many are exploring tools that rely on text interpretation, video analysis, or structured data. These approaches often powered by large language models or computer vision focus on what is said or what is seen. While powerful in controlled settings, these tools often face limitations in real-world environments, especially in linguistically and culturally diverse regions. Differences in language, dialect, nonverbal norms, and communication styles can reduce model accuracy, introduce bias, or require extensive customization. This makes it challenging to deploy such tools at scale across varied populations.

What sets behavioural voice technology apart is that it focuses on how people speak, not what they say. Rather than analysing content, the platform identifies behavioural markers from vocal parameters like tone and pace. This approach is language-independent, culturally neutral, and capable of generating real-time insights from short voice samples. In a diverse region like Southeast Asia, where multilingual environments are common, this offers a significant operational advantage.

In discussion with Haf Jacobson, CEO of Voicesense, he said that the company’s platform is deployed in real-world settings globally, including with governments and enterprises throughout the United States, Europe, and Asia. In Singapore, the company is currently involved in a multi-phase pilot with the Prison Service and Home Team Psychology Division. These programs aim to support the early detection of stress, burnout, or relapse risk through passive, voice-based interaction. The technology provides timely, data-driven feedback without requiring manual monitoring or invasive testing.

Public agencies expressed particular interest in the platform’s ability to deliver actionable insights, while maintaining strict data protection standards. Voicesense uses a privacy-by-design model, processing only non-identifiable acoustic features and storing no voice data beyond short processing windows. The platform can be deployed in on-premise or hybrid environments, making it suitable for regulated sectors such as corrections, healthcare, and defence.

Beyond government, behavioural AI is gaining ground in finance and healthcare. Banks and insurers are using behavioural indicators to strengthen fraud detection, customer profiling, and churn prediction. In healthcare, short voice samples can help track patient well-being remotely, flagging subtle behavioural changes over time. This offers clinicians a lightweight, continuous monitoring tool that complements their own observations.

Across Southeast Asia, interest in behavioural AI continues to grow. The demand for low-resource, explainable AI tools that work across diverse populations is driving adoption in countries like Indonesia and the Philippines. These countries are piloting the technology in areas such as employee well-being, financial inclusion, and telemedicine.

The success of these applications often hinges on partnerships. Voicesense has adopted a collaborative model, working with local pilot programs and innovation hubs such as Hatch to tailor its tools to specific agency workflows and cultural needs. The company claims its partner-first strategy has proven effective in translating policy goals into measurable impact.

As behavioural AI becomes more mainstream, institutions are focusing not just on innovation, but on responsible implementation. Solutions must be transparent, secure, and designed to assist rather than replace human judgment.

The conversation at ACCOP showed that AI is no longer just about algorithms. It is about empathy, context, and understanding people more deeply. Behavioural voice analytics may prove to be one of the most human uses of AI yet, helping institutions make better decisions by listening in a new way.