Why intent prediction needs more than an LLM

Summary

Traditional large language models rely on next-token prediction trained over broad text sequences, excelling at conversational synthesis and code generation while struggling with forecasting future user behavior. Predicting intent inherently requires modeling decision-making under conditions of uncertainty, a capability not directly produced by general text pretraining. Dedicated behavioral foundation models address this limitation by training representations directly on proprietary, sensitive behavioral data, such as anonymous browsing sessions and product identifiers. By learning base representations of behavior, these models generalize to unseen predictive tasks, such as optimizing ad campaigns for products absent from the original training dataset. Implementing these behavioral models also necessitates privacy-centric machine learning techniques, ranging from differential privacy concepts to homomorphic machine learning.

Context

Next-token prediction in large language models relies on text-based pretraining that lacks the inductive bias needed for forecasting, decision-making under uncertainty, and user intent prediction.

Approach / What changed

Building dedicated behavioral foundation models trained directly on proprietary and sensitive interaction data to learn representations that generalize across downstream predictive tasks.

Takeaways

• Standard LLM inductive bias from next-token text training does not naturally equip models for decision-making under uncertainty or forecasting user intent.
• Behavioral foundation models learn generalized representations from proprietary, non-text interaction signals to predict outcomes for novel products and campaigns.
• Differential privacy and k-anonymity are typically used for data analysis and querying rather than directly inside model training workflows.

Related reading

• Stack Overflow · Aug 20, 2026

  From PHP to team lead of agents: rethinking judgment, review, and data with Google's Andi Gutmans (Part 1)

  Software development is experiencing an architectural shift as artificial intelligence agents transition individual contributors into orchestrators and reviewers of automated workflows. Rather than spending effort on manual coding, developers increasingly focus on defining system architecture, guiding agent execution, and managing operational risk through tiered human and agent review loops. The primary bottleneck for agentic systems lies in preparing organizational data so that agents can effectively reason over enterprise knowledge. To address this challenge, Google published an open knowledge format and is adopting agent-driven semantic ontology generation to infer relationships across messy data estates. This evolution transforms ontology building from manual human toil into automated generation, allowing engineers to focus on high-value data curation and strategic system design.

  Eira May

• Stack Overflow · Jun 18, 2026