FIFTEEN YEARS
OF CAUSALITY.
Divera did not start as a product. It started in 2011 as a structural question: how do you model individual consumer behavior when marketers have almost no data, and the data they do have is noisy, biased, and self-reported?
From a 2011 question to a neural decision engine.
Most AI systems are built for environments with massive, clean datasets. Marketing is the opposite: sparse panels, survey bias, and a handful of transactions per person. Traditional machine learning collapses in this regime.
Divera's founder spent the 2011–2014 window working almost exclusively on that gap—how to model individual consumer causality when the input signal is thin, and how to keep the model honest enough that a strategy team could actually defend its recommendations.
Everything in the current platform—individual mind models, biological signal, explainable outputs, simulation-first method—traces back to decisions made in those four years.
-
2011
The question is formed.
Founder begins a systematic review of consumer decision modeling under data scarcity—identifying that the market was buying averages and calling it insight.
-
2012–2013
Individual causality, not segment averages.
Early prototypes shift the unit of analysis from segment to individual: one decision engine per person, grounded in causality rather than correlation.
-
2014
Neuroscience enters the stack.
Biological signal is added as a first-class input—capturing the subconscious component that surveys structurally miss—paired with neural network architectures suited to small, structured data.
-
2015
First long-horizon enterprise partnerships.
The method graduates from research to deployment alongside enterprise strategy teams—including a multi-year program with Beko that continues today (30+ projects and counting).
-
2016–2023
Simulation becomes the layer above the model.
Mind models are combined into bottom-up simulations where millions of individual decisions aggregate into market-level outcomes—not estimated, but composed from the agents' own actions.
-
2024–Now
Explainable, one-shot, deployable.
Current platform emphasizes explainable outputs (XAI), one-shot learning so teams can model new categories without huge training sets, and long-term implementation support rather than one-off licenses.
“We didn't set out to build another AI tool. We set out to model the one thing marketing keeps pretending it understands: why a specific person makes a specific choice.”
Polish HQ. EU strategy, embedded.
Headquartered in Poland, Divera runs AI and neuroscience-based strategy infrastructure for enterprises across Europe and beyond—embedded with client strategy teams, not shipped as a standalone license.