Worldwide spending on AI is forecast to reach roughly $2.5 trillion in 2026, an increase of more than 40% in a single year. In the same window, MIT’s NANDA initiative reports that about 95% of enterprise generative AI pilots and projects deliver no measurable P&L impact. Hold those two numbers next to each other. The largest technology build-out in history is producing, for most of the companies paying for it, almost no change in what people actually do.
That is not a model problem. The models are extraordinary and getting better every quarter. It is a motivation problem, and it points at a layer the AI stack does not yet have, the layer we built as the Octalysis AI Engagement Engine.
Virtually every production AI system is optimized to model behavior. It records what a person clicked, bought, watched, asked, and abandoned, and predicts the next action with accuracy. The sophisticated ones go a step further into propensity and intent: a churn score, a likelihood-to-convert. What none of them yet has is a standardized, interpretable model of motivation, the layer that explains why the behavior happened.
Not knowing this is a real gap that needs to be filled. Say that two users in an app quit the experience at the same step. One was bored. One was anxious. A behavior model sees a single churn signal and sends both the same nudge, saving one and pushing the other out the door.
These people may even have the same interests, similar friend circles, similar urban and educational backgrounds, and may have clicked on similar images before. It doesn’t mean that AI knows nothing; it just doesn’t know what the motivation was at that moment for that user. And that’s exactly what the Octalysis Group has built and patented the method behind it. It is the Octalysis AI Engagement Engine. It is not the only way to think about motivation, but it is the most tested one we know how to put into software.
What follows is the case for why it is the missing piece, how it plugs in, the evidence that it works, who it is for, and why it could only have come from us.
The Missing Layer: Motivation Intelligence
Recommendation engines were the last big layer added to the AI stack. They answer one question well: what should this user see or do next? Two decades of behavioral data and ever-larger models have made next-action prediction close to solved.
A motivation layer answers a different one: why is this user doing this, and what would move them? Retention, loyalty, and real engagement turn on that question, and the stack has no standard way to answer it. Propensity scores and lookalike models get close. They still guess from surface behavior without naming the drive underneath, which is why opposite motivations end up with the same label.
But it’s not perfect at all. That’s why you still get posts or images suggested by the algorithms of Instagram, Facebook, or X. The algorithm keeps trying to fire new content at you in the hope of finding more use patterns so that it can predict what the next content should be. However, it’s highly dependent on curation and doesn’t inform you exactly about my motivational state at that moment or in similar moments.
The Octalysis AI Engagememnt Engine fills the gap and creates what we call motivation intelligence: an interpretable layer over behavioral data that estimates why people act. The Engine is built on a complete model of human motivation rather than a handful of personas. The Octalysis Framework’s eight Core Drives are the basis of the engine to map every human motivation possible. are a working vocabulary, refined over twenty years, that teams can actually design against.
Why Now
For years now, AI capability was the moat. And still now, the company with the better model wins, so that is where the money went. That era is ending.
Models are converging. Open-weight models trail the frontier by months, not years, and most users cannot tell which model answered them. Once everyone can reach a great model, the model stops being the differentiator. One question is left, the one the industry keeps avoiding: when nothing forces the choice, whose product do people come back to?
That question is behavioral, not technical. We think the next defensible layer in AI will not be a bigger model. It will be a deeper read of the people the model serves. The companies that get there first will hold an advantage that better models alone cannot erode, because a competitor can match your model in a quarter and cannot match twenty years of understanding why humans engage.
The economics make the timing sharper still. With AI spend approaching $2.5 trillion a year and the overwhelming majority of deployments failing to change behavior, the gap between what is being spent and what is being returned is now measured in hundreds of billions of dollars. That gap is the market for motivation inference. Closing even a fraction of it is the difference between an AI investment that compounds and one that quietly gets written off.
What the AI Engagement Engine Does
The Engine is a motivation inference layer that sits on top of the behavioral data a product already collects and turns it into a live, evolving read of what drives each individual user. We will describe what it takes in and what it produces. We will not describe how the inference works, because that is the patented core.
The inputs are ordinary product events. When someone returns. What they ask for. What they customize. Which features they keep using. What they make and keep. How they treat other people in the system. No survey, no quiz, no interrupting question. People show what drives them through what they do, and the Engine is built to read it.
The output is a profile across the eight Core Drives of the Octalysis Framework, the model of motivation Yu-kai Chou created and we have applied across hundreds of engagements. For each user, the Engine estimates how strongly they are driven right now by meaning, accomplishment, creativity, ownership, social influence, scarcity, curiosity, or the avoidance of loss. The more actions a person takes in an experience (click a CTA, add a friend, buy an item etc), the sharper the read.
The profile is not a fixed label. Motivation is not a personality type. It moves as someone goes from first use to habit and mastery. The first time you interact with Facebook, your skills and motivations are very different than the 100th time that you interact with Facebook. The Engine tracks that shift, so the read stays current instead of freezing a user as whoever they were on day one.
With that read, a product can finally do what personalization always promised. It can shape the challenge it sets, the social cues it shows, the rewards it offers, and the way it speaks, around the actual person rather than a demographic average. The same assistant can stretch the user who wants mastery and reassure the user who wants to belong, without asking either of them anything.
How the Engine Plugs Into Your Stack
The Engine runs beside your existing AI and data infrastructure. It is not a rip-and-replace, and it is not a consulting project. The interfaces are deliberately plain.
It takes the event streams you already produce: logins, return cadence, the actions and prompts a user makes, configuration choices, the features they revisit, and social signals such as sharing. Context like device or cohort sharpens the read. You add no surveys and little or no new instrumentation.
It returns a per-user, per-phase distribution across the eight Core Drives, through an API and a dashboard. Product, growth, and AI teams read those signals and adapt flows, messaging, rewards, and agents to each user’s profile in near real time. Because it is a layer rather than a platform you migrate onto, you can run it as hosted software, in a private cloud, or as a licensed model inside your own environment.
Twenty Years of Motivation Data
Here is the fair challenge from a platform exec or an investor: couldn’t a well-funded team point a behavioral-science group and a few hundred million events at this and build the same thing? They could build a model. They could not build what makes the model mean something. The moat is three layers, and only one is code.
The first is the ontology. The Octalysis Framework, created by Yu-kai Chou and one of the most widely used frameworks in behavioral design, gives the Engine a named, designer-ready vocabulary of eight Core Drives, tested across hundreds of products and cultures. A rival’s latent factors might cluster behavior, but a product team cannot design against numbers it cannot read. Named drives are something a person can build with.
The second is the interpretive corpus. Across more than 175 engagements, from global banks to Fortune 500 brands, we have built a private library of behavior-to-motivation mappings: which drives produce compliance, which produce commitment, how the mix shifts through an experience, and what happens when a drive is fed or starved. That corpus grounds the Engine’s reads, and it exists nowhere a competitor can buy or scrape.
The third is the codified mechanism, the patent-pending method that turns a product’s raw events into that shared motivation space. A frontier lab can match the model. It will struggle to match the ontology, the corpus, and twenty years of knowing what the output means.
The Proof: Designed Motivation Wins, With Money Attached
The idea behind the Engine is not a theory. We have spent twenty years showing, in production, that designed motivation beats raw capability, and that the gap is large enough to move the numbers a business reports.
A Procter & Gamble distributor let its sales force keep an efficient CRM or switch to an Octalysis-designed experience built around intrinsic motivation. Given a free choice, the workforce adopted the designed version and stayed with it for years. Over the same period the distributor reported revenue up 28.6% and voluntary adoption above 99%, and the program won a Best Gamification Project award. Revenue moves on many things, so adoption is the cleaner signal: capability was identical, the design was the only difference, and people chose it.
CAIXA Econômica Federal, one of Brazil’s largest banks, rebuilt the daily work of more than 80,000 people around intrinsic Core Drives instead of top-down targets. During the program, recurring net income rose from 8.6 billion to 12.7 billion reais in a year, on the order of a billion dollars, with Octalysis design among the contributing factors, and the bank moved from second to first. These are our own engagements, not peer-reviewed studies. We ran no controlled trials and will not claim Octalysis design was the only cause. What we can show is that, given a free choice between two functionally identical tools, people chose and kept the one built around intrinsic motivation.
That is the work the Engine productizes. For twenty years it took our people in the room, mapping a workforce or a customer base by hand. The Engine performs the same read continuously, per user, inside the product itself.
What the Engine does not do is worth saying plainly. It does not read minds. It does not guarantee lift on its own. It does not replace product design. It infers likely motivational states from behavior, and it pays off when teams act on those reads and rebuild the experience around them.
Who This Is For
Two kinds of organization gain the most, and they want different things from the same capability.
The first is the platform companies whose business runs on behavioral models. The personalization, growth, and ranking teams at the largest AI and consumer-tech companies are the best in the world at modeling what people do, and weak at modeling why, because behavior is silent on cause. For them the Engine turns a behavior model into a motivation model: the difference between predicting the next click and understanding the person who clicks (and why they click, or don’t click). It adds to what they have already built, because it reads data they already hold. We have written about how this plays out in consumer products and the workforce; the Engine is the layer under both.
The second is the investors watching where AI value is heading. As capability commoditizes, the durable returns go to whoever owns the layer capability cannot replicate. A model of human motivation built over twenty years and protected by patent is that kind of asset. For an investor, the Engine is a bet that the next phase of AI is won on understanding people, not on raw compute.
The Next Layer in AI
The first wave of AI was about capability, and that race will be settled in the not so distant future. Models can do remarkable things, and soon every serious company will have one nearly as good as everyone else’s. The next wave is about people: whether a product understands the human in front of it well enough that they choose it, go deeper, and stay.
That does not come from a larger model. It comes from a working understanding of human motivation, applied continuously, inside the product. It is what we have spent twenty years learning to do by hand, and it is what the Octalysis AI Engagement Engine is built to do automatically.
For platform and product teams: if engagement is flat despite a capable model, the Engine is the layer you are missing, and it reads data you already hold. We are working with a small number of partners to put it into production.
For investors: if you think the lasting returns in AI go to whoever owns the layer capability cannot copy, this is that layer, built on twenty years of work and protected by patent. We are talking with a few investors who want to back motivation intelligence rather than the next model.
Either way, start a conversation with us. The companies that win the next phase of AI will be the ones people keep coming back to once the novelty wears off. Building that pull is the work we have done for twenty years.
Joris Beerda is CEO and Co-Founder of The Octalysis Group, the behavioral design consultancy behind engagement programs for Microsoft, P&G, Porsche, Booking.com, and CAIXA Econômica Federal. He is co-inventor of the patent-pending Octalysis AI Engagement Engine, alongside Octalysis Framework creator Yu-kai Chou.
Frequently Asked Questions
What is the Octalysis AI Engagement Engine?
It is a patent-pending motivation inference layer for AI. The Engine reads the behavioral signals a product already collects and infers what drives each user, expressed as a live read across the eight Core Drives of the Octalysis Framework. A recommendation engine predicts what a user will do next. The Engine estimates why they do it and what would keep them engaged, so a product can adapt to the actual person without sending a survey.
How is motivation inference different from personalization?
Conventional personalization decides what to show a user based on what similar users did. Motivation inference decides what the experience should feel like for this specific user, based on the drives behind their behavior. Two people can take the identical action for opposite reasons, so behavior-based personalization cannot tell them apart. Motivation inference can.
Does the Engine require surveys or extra data collection?
No. It is built to avoid surveys, which interrupt the experience and produce answers that decay fast. The Engine reads motivation from behavioral signals most products already generate: return cadence, the kinds of actions taken, what users customize, and how they interact with others.
Who is the Engine for?
Two kinds of organization. Platform and product companies whose value depends on engagement and retention, for whom the Engine adds a motivation layer on top of their existing behavioral models. And investors looking for the durable layer in AI as raw model capability commoditizes.
Is the Octalysis AI Engagement Engine available now?
It is rolling out with a small number of partners rather than generally available. We are integrating it with selected platform and product teams and speaking with investors who want to back motivation intelligence. If you want to be early, start a conversation with us.