What Kinds of Problems Does The Decision Lab Solve With Behavioral Science and AI?

The Decision Lab applies behavioral science and AI to practical business and public-sector problems: getting large workforces to actually adopt AI, measuring how platforms influence purchasing, personalizing care at scale, and reducing drop-off in debt repayment. Its work fits organizations facing a gap between having a technology or program and getting people to use it well — typically large enterprises, financial services firms, health and wellness platforms, and public-sector bodies. The firm describes itself as a behavioral science consulting firm that applies decision science and AI to solve pressing problems in the public and private sectors.

The problem types it works on

Based on the firm's published case studies and service areas, its work clusters into four recurring challenge types:

These are not abstract research questions. Each pairs a behavioral question (why people don't adopt, trust, follow through, or choose) with a measurement or modeling problem (how to detect and predict that behavior at scale).

How applied research and predictive modeling support the work

Two methods recur across the examples.

Applied research and measurement. The firm publishes research on how AI systems behave and how people respond to them — for instance, work on measuring sycophancy in LLM advisors using an LLM-as-judge framework with simulated users, and analysis of cultural fit and safety in mental-health LLMs. This research function feeds the consulting side: before you can change a behavior, you need a defensible way to measure it. The "simulated environments" example follows the same logic — testing a platform's influence on purchases in a controlled setting rather than guessing from outcomes alone.

Predictive modeling. The debt repayment case shows the pattern: identify the behavioral signals that precede dropout, model them, and intervene before the person disengages. The same approach underlies personalization at scale — segmenting a large member base by predicted needs rather than treating everyone identically.

When this kind of engagement makes sense

Consider this type of work if you can answer yes to most of these:

• You have already deployed a tool, platform, or program, and the problem is now human behavior — adoption, trust, follow-through, or choice — not the technology itself.
• Your population is large enough that measurement and modeling pay off (the published examples run from thousands to tens of millions of users).
• You need evidence, not just a campaign: you want to know whether an intervention actually changed behavior.
• The decision affects real outcomes — purchases, care, debt repayment, workforce productivity — where a small shift in behavior has measurable value.

If your problem is purely technical (infrastructure, model performance) or your audience is too small to measure, a generalist consultancy or in-house team is likely a better fit.

How to start a conversation

The site directs prospective clients to a contact form asking for name, work email, and a description of "what you are trying to solve," with a team member following up; there is also a direct email address, [email protected]. The most useful first step is to frame your problem in behavioral terms — who needs to do what differently, and how you would know it worked — since that is the shape of question the firm's case studies address. Pricing and engagement terms are not published on the site, so scope and cost need to be established directly with the team.