"We're trying to mimic the human brain," Eric Pritchett, President/COO of Terzo, stated in a recent discussion about preparing IT for the advent of AI agents. Pritchett, speaking with an interviewer, elaborated on how modern IT infrastructures need to evolve to become "AI-ready," drawing a compelling analogy between the human brain's processing capabilities and the future architecture of AI-driven systems.
The core of Pritchett's argument centers on a fundamental shift in how we approach AI integration within organizations. He illustrated this by sketching a conceptual model of the human brain, dividing it into three primary regions: the lower brain, the mid-brain, and the upper brain. The lower brain, he explained, handles primitive functions, processing raw data and generating basic responses. This is akin to how early AI systems, and even current large language models, process vast amounts of information from the internet to generate text or images. "AI swallows the internet and processes it," Pritchett described, highlighting the sheer volume of data these models consume.
However, he quickly pivoted to the limitations of this approach for enterprise-level AI. "The data we care about inside an organization is very different than all the data that's on the internet," Pritchett emphasized. This distinction is crucial. While the internet provides a broad, often unfiltered, dataset, enterprise data is specific, contextual, and often proprietary. The challenge, therefore, lies in bridging the gap between the broad, generalized capabilities of current AI models and the specific, nuanced needs of a business.
Pritchett articulated this challenge by contrasting the current AI paradigm with a more integrated, human-like approach. He noted that while current AI models "swallow the internet and process it," leading to models with a "very heavy overlap with AI," this doesn't necessarily translate to effective internal enterprise applications. The failure rate for AI initiatives, he suggested, is high precisely because organizations often try to "jam AI into the existing enterprise" without re-architecting their foundational systems.
The key insight here is the need for a more sophisticated AI architecture, one that mirrors the brain's ability to process, store, and selectively recall information. Pritchett highlighted that just as the human brain has distinct regions for processing sensory input, memory, and executive functions, IT infrastructures need to be similarly segmented and interconnected. He proposed a three-tiered approach: applications, data, and network. "We have applications, we have data, we have network," he stated, laying the groundwork for a more structured AI integration.