Develop Predictive Models for Additional Purchase Behaviors

Design and implement machine learning models that predict additional customer purchase behaviors, including:

• Up-sell opportunities
• Promotion responsiveness
• Customer churn risk

These models will complement existing recommendation and timing systems to create a more complete behavioral prediction layer.

Establish an LLM Architecture and Utilization Strategy

Design and implement a robust LLM architecture and integration strategy to enhance the company’s capabilities as an AI-driven technology and research organization. This includes identifying high-impact use cases, defining infrastructure and orchestration patterns, and integrating LLMs into both internal workflows and product features.

Develop Theory-of-Mind–Based End-Customer Engagement Modeling

Initiate the development of Theory-of-Mind (ToM)–based models for long-term engagement prediction of end-customers interacting with our tenants (B2B clients). The objective is to model how users’ beliefs, preferences, and contextual signals influence their engagement and purchasing behavior over time. This effort requires additional behavioral data collection and evaluation of state-of-the-art modeling approaches.

Build an Agentic Framework for Continuous System Improvement

Develop an agentic improvement framework that enables AI systems to learn, adapt, and self-improve over time.

The framework will initially focus on existing solutions such as:

• Metadata enrichment
• Substitute recommendation
• Cross-sell systems

The long-term objective is to create a self-optimizing AI ecosystem where agents monitor performance, propose improvements, and adapt models or pipelines accordingly.