The article discusses a significant shift in enterprise AI adoption, moving away from opaque 'black box' models towards 'grounded' or 'retrieval-augmented generation' (RAG) models. It explains that while powerful, traditional large language models (LLMs) often produce hallucinations, lack company-specific knowledge, and cannot cite their sources, making them risky for high-stakes business decisions. Grounded models address …

A new language model architecture called Bolmo has been introduced, designed to enable efficient training at the byte level without compromising output quality. Traditional models typically use subword tokenization, which can be inefficient and limit the model's ability to handle diverse data formats like code or non-Latin scripts. Bolmo addresses this by employing a novel, …

The article argues that the rise of 'agentic AI'—autonomous systems capable of planning and executing complex tasks—requires a fundamentally new category of customer data. Traditional customer data, often historical and static, is insufficient for these proactive agents. The piece introduces the concept of 'intent data,' which captures a user's real-time goals, context, and desired outcomes. …