Most companies are still at the pilot project or employee assistance stage, according to survey results presented in a content piece produced by Insights, the custom content division of MIT Technology Review. For Prukalpa Sankar, co-founder of Atlan, the main shortfall lies less in the models than in the context in which they are used.
A Gap Between Capabilities and Usage
According to these results, 38% of respondents indicate that their company is mainly conducting pilot projects in AI, while 35% use it to help employees in their work, with tools such as copilots or conversational assistants. Only 12% of companies have AI agents in production, and 3% claim to use AI to accomplish the majority of work in key processes.
“There is a large gap between what the models are capable of doing and how they are actually useful within organizations,” says Prukalpa Sankar. She adds: “For every concrete use case, general artificial intelligence is already there. It’s just not really useful in most companies today.”
The Context Layer
According to her, the missing piece is context, meaning the AI’s ability to properly understand a company. She describes a “context layer” as an architectural infrastructure placed between an organization’s business systems, for example systems of record and data, and its AI platforms. This layer translates enterprise information into a structured, machine-actionable context intended for AI agents or models, and allows knowledge to accumulate.
“Without an AI context layer, it’s almost like hiring the world’s brightest intern, blindfolding them, and throwing them into your organization,” says Prukalpa Sankar.
This content was written by humans, according to MIT Technology Review, and was not produced by its editorial staff. Any AI tools used were limited to production processes, under human supervision.
