Impetus builds an operational framework to bridge AI’s ‘context gap’

Impetus Technologies is betting that the big problem with business AI isn’t the model. It’s the gap between what the model knows and what a company actually needs it to know. Deepak Khosla, the company’s chief growth officer and head of AI, argued that the missing piece is context: where the data lives, how the business works, and what the agents are supposed to do with it.

That idea sits at the center of Impetus’ pitch for enterprise AI. The company says it has built an operational framework to test and govern the data that AI agents use, so they can do more than answer prompts in a generic way. Khosla described the approach as CEDL, short for Context Engineering Delivery Lifecycle, a method that creates context, engineers it, learns from what happens in practice, and feeds those signals back into the system.

The framework leans on the old but suddenly fashionable enterprise habits: modernizing legacy systems, building knowledge graphs, defining an ontology layer for business tools, and creating memory. And memory, in Khosla’s telling, matters a lot. If an AI model learns the wrong thing, that mistake can stick and shape later actions.

Impetus is packaging this thinking through its Leap AI suite, which is meant to help move AI into production. The company’s pitch is that it can build the semantic layer and knowledge layer while it is also migrating and modernizing a customer’s data platform. In other words, the agents are not supposed to arrive at a company as blank, overconfident interns. They get briefed.

Khosla made the case on theCUBE, SiliconANGLE Media’s livestreaming studio, in a sponsored segment. The broader argument is a familiar one in enterprise AI: the model may be smart, but the organization still has to teach it how business really works.