Behavioral science for AI.
Models get validated. Policies get written. Then real people, under real pressure, use the system, ignore it, or over-trust it in ways nobody designed for. Behavieural helps organizations understand and design for that, with evidence.
-
90+
peer-reviewed articles in top academic journals. (Google Scholar)
-
Top 2%
of academic researchers worldwide for our Founder and Principal Strategist. (SCINET)
-
4
continents across which over dozens of engagements have been delivered: Asia, Australia, Europe, and North America.
-
EU
-recognized expert in the areas of AI adoption and behavioral science. (EPRS/EVAL/SER/25/008CEI)
-
TSI
Advisor to the Trust and Safety Institute, dedicated to building, assessing, and demonstrating trustworthy AI agents. (TSI)
Behavieural applies behavioral science to how people trust, use and oversee AI, and turns that evidence into adoption, governance, and oversight that hold up in practice. The work draws on research in psychology and decision science, and it is offered as consulting, advisory, speaking and training.
-
AI systems rarely succeed or fail on technology alone. They succeed when people understand them, rely on them to the right degree, and can step in when something goes wrong. They fail when people avoid them, work around them, or defer to them without checking. These are behavioral patterns, and they can be measured and changed.
-
-
Most organizations track what is easy to count, such as usage, accuracy and compliance. Those measures describe the system but not the people around it: what they believe, what they do when no one is watching, and where policy and practice diverge. That is where adoption stalls, governance weakens, and oversight turns into a formality.
Latest Thinking on AI & More
-
![The Confidence Gap]()
The Confidence Gap
An original survey of professionals who evaluate or oversee AI in healthcare, finance and manufacturing. Where leaders and users diverge, and what to do about it.
-
![Governing Agentic AI]()
Governing Agentic AI
Five behavioral mechanisms that shape the human–agent relationship, as a behavioral threat layer beneath the OWASP Top 10 for Agentic AI.
-
![AI Trust Axis]()
AI Trust Axis
The AI Trust Axis is a behavioral framework for measuring how people come to trust and rely on AI systems. It breaks the human response into five dimensions, each tied to a recognizable failure such as overreliance, avoidance or performative oversight, and it produces scores for deployed systems and for readiness before deployment.
-
![From Indicator to Claim]()
From Indicator to Claim
This white paper offers a method for reviewing AI standards, and the evidence offered under them, when the claim concerns human behavior. It shows how to read a requirement as a validity argument, which threats are specific to behavioral evidence, and how much evidence a given level of risk warrants. The method is unvalidated, and the paper recommends piloting it before it is relied on.
Where Are You With AI?
-
Deploying AI
Tools are in place and adoption has stalled, or it works in one team and won’t spread. I diagnose why people aren’t relying on the system and design the changes that make use last.
-
Selling AI
Customers show low usage, workarounds or misplaced reliance, and the risk lands on you. I locate where the breakdown happens between your product and its users.
-
Governing AI
Policies exist, but you can’t show that people follow them or that human oversight is real. I measure whether governance and oversight work in practice and help build ones that do.
-
Not Working with AI
Brand trust, crisis response and customer loyalty depend on the same behavioral mechanisms. The Great Canadian Brand Index, the STORM Framework, and related tools apply them.
Examples of Our Work
-
Fixing the Human Side of the Model You Can't Touch: A Buyer-Side Trust Calibration Engagement in Commercial Lending
A mid-market lender's off-the-shelf AI credit scoring tool was technically sound but underwriters couldn't explain its declines, leading staff to quietly build a shadow manual review process alongside it. A behavioral diagnostic found the real gap wasn't the model but the bank's missing discretion policy, explanation vocabulary, and appeal feedback loop—building all three internally cut shadow reviews to near zero and appeal reversals from 34% to 19%.
-
Reducing Member Churn for an Australian Yoga Studio
A mid-sized yoga studio in Australia struggled with high customer churn despite strong reviews and a loyal community. A behavioral science analysis of existing membership and attendance data revealed that members were not leaving because of dissatisfaction, but because of hidden behavioural barriers—including early-stage friction, choice overload, weak habit formation, and limited visibility of progress. By redesigning the member journey around these insights, the studio reduced churn from 38% to 22%, increased attendance consistency, and extended membership duration. The case demonstrates how behavioral science can uncover the psychological factors that drive customer loyalty and retention.
-
When Adoption Stalls: Diagnosing AI Trust Failure in a Mid-Market Insurance Brokerage
A mid-market insurance brokerage struggled to drive adoption of an AI underwriting assistant despite strong technical performance and seamless implementation. A behavioral audit revealed that low usage was driven by trust barriers, including unclear reasoning, social risk around overrides, and the outsized impact of early errors on perceptions of reliability. By redesigning explanations, normalizing human judgment, and rebuilding confidence through targeted interventions, the brokerage tripled senior broker adoption, reduced shadow workflows, and achieved a 22% improvement in case-handling time. The case demonstrates that successful AI adoption requires behavioral trust architecture—not just technical capability.
-
Strengthening Trust in an Autonomous Financial‑Planning Tool
A major German financial-services provider struggled to drive adoption of an autonomous financial-planning tool despite strong initial engagement. A behavioral diagnostic revealed that users were not limited by comprehension or onboarding issues, but by trust barriers around control, explainability, and sensitivity to perceived errors. By redesigning the experience to strengthen user agency, improve explanations, and introduce gradual delegation, the provider increased delegation rates by 38%, reduced onboarding drop-off, and helped customers move from passive monitoring to active reliance on AI. The case demonstrates that autonomous AI adoption depends on building psychological trust, not just technical confidence.
-
How a Fairer Loyalty Program Increased Carbon Offset Participation
A major airline group struggled to increase participation in its carbon-offset program despite growing consumer interest in sustainable travel. A behavioral science analysis revealed that engagement depended not only on awareness, but on whether members perceived the loyalty program as fair, transparent, and attainable. By understanding how fairness and reciprocity influence prosocial behaviour, the airline identified ways to redesign the member experience and increase participation. The case demonstrates how behavioral science can help organizations strengthen loyalty by aligning customer motivation with broader brand goals.
-
How a Website Chatbot Earned Customer Trust Through Psychological Insight
A major consumer service provider struggled to drive adoption of a high-performing support chatbot despite strong technical capabilities. A behavioral audit revealed that customers were not avoiding the tool because of functionality, but because of trust barriers related to uncertainty, perceived competence, and fear of losing access to human support. By redesigning the experience around clarity, confidence, and user control, the provider increased chatbot engagement by 31% and reduced unnecessary escalations by 27%. The case demonstrates how behavioral science can help organizations build trust in technology and unlock its intended value.