The foundation model for transactional behavior.
We are building a new class of foundation models for sequential human behavior in retail, payments, and finance. Large Behavioral Models (LBMs) that (1) learn directly from transactional behavior, (2) generalize across users, merchants, and geographies, (3) transfer to downstream applications including personalization, credit scoring, and fraud detection.
Ultimately, we're pursuing general behavioral intelligence: what we think of as the AGI of human behavior in the digital economy.
Our research
2026-02-05
Large Behavioral Models: A Foundation Model Paradigm for Human Actions
We introduce Large Behavioral Models (LBMs), foundation models trained on chronological sequences of human actions. BehaviorGPT, a frontier LBM, delivers double-digit sales uplift for retailers and payment companies.
2025-07-15
BehaviorGPT: aesthetic purchasing preferences from behavior
We learn aesthetic preferences from purchasing and engagement sequences (how people discover, save, and acquire art and design) rather than from static labels alone. Trained on 215 billion human interactions across major art and design platforms.
2025-05-15
A Foundation Model for Consumption, Transactions, and Actions
We scaled language modeling to grocery consumption (hundreds of millions of online actions and billions of offline purchases), treating each shopper's history as a sequence to predict future events: the inception of BehaviorGPT before it extended to payments and retail more broadly.
Unbox in action
Methodology
- Modeled customer behavior as time-series sequences (sessions + transactions)
- Scored users in real time to predict intent and next purchase
- Powered personalized search ranking, recommendations, and dynamic category pages
Customer profile
- Large online art retailer
- Millions of active shoppers
- Competing directly with best-in-class search providers
“This is the largest A/B test impact we’ve seen in our company’s history.”
Staff Engineer, ClientMethodology
- Learned shopper intent from checkout sequences
- Scored each checkout in real time to predict drop-off risk and likelihood to complete
- Ranked payment options, messaging, and offers to keep checkout smooth and improve completion
Customer profile
- Leading fintech platform
- 3.4M daily transactions
- 850,000+ retail partners
What Unbox can do for you
Team
Join the teamRickard Brüel Gabrielsson
“Current AI models understand language, but they struggle to understand agency. We are building the foundational layer for how machines interpret human intent, not by what people say, but by the sequence of actions they take. This is a once-in-a-generation opportunity to define an entirely new category of AI, and we are looking for the people who want to build that history with us.”
Gunnar Carlsson
“Every interaction creates a signal, but most enterprises are only listening to the noise. We founded Unbox to turn fragmented behavioral data into a coherent narrative, giving businesses the ability to predict the next move. Our vision is simple: one universal model for human behavior that powers every application where understanding people matters.”