Enterprise AI: Fewer Than One-Third of Organizations See Substantial Business Impact
Fewer than one-third of organizations see substantial business impact from their AI initiatives. Faisal Saeed, the Founder & CEO at Promptev Inc, a leading AI solutions provider, says the problem with enterprise AI isn't the technology itself, but everything around it. In a recent article, Saeed cited a 2013 incident where a major U.S. airline experienced a widespread outage in its reservation system, highlighting weaknesses in testing, rollback preparedness, and operational resilience. The disruption resulted in strained customer confidence during a critical period for the airline.
Saeed notes that most companies have poured their energy into asking which AI model is right for them, but this is a reasonable question - and it's the wrong one to obsess over. The real issue is the operational backbone that supports the AI model. According to McKinsey's State of AI research, most companies struggle with context, reliability, and governance, which are critical to deploying AI successfully. Context is a significant issue in enterprise AI because AI models are often trained on enormous amounts of data, but not the specific data relevant to the organization. As a result, the AI may make decisions based on incomplete or outdated information, leading to inconsistent results.
Saeed cites the example of an AI system that is trained on a dataset from the US, but is deployed in a European market, where the data may not be relevant or up-to-date.
Reliability is another challenge in enterprise AI. Without a way to monitor and test the AI's behavior, organizations may receive inconsistent results from the same AI system. This can lead to costly errors, such as loan approvals or compliance risks. In regulated industries, this drift can carry legal exposure, while in consumer-facing products, it erodes trust. Organizations need a reliable way to monitor and test the AI's behavior to ensure consistent results.
Governance is also a critical issue in enterprise AI. Most AI deployments today are built to produce outputs, not explanations. When something goes wrong, there is no clean record of what the AI was told, what it decided, or why. This can make it difficult for organizations to explain their decisions to regulators, customers, or even their own boards. Governance documentation exists somewhere in a shared drive, but it's disconnected from what the AI is actually doing day-to-day.
Faisal Saeed argues that the issue is structural, and most AI systems were designed to be powerful, not managed. The team that built the model is different from the team that handles compliance, and the data pipeline was set up quickly and has not been properly maintained. As a result, governance documentation exists somewhere in a shared drive but is disconnected from what the AI is actually doing day-to-day. In other words, the operational backbone that supports the AI model is often inadequate.
A shift is happening among the enterprises getting AI right, however. They are no longer asking which AI model to use as their first question, but rather how to build an environment where AI can actually be trusted. This means treating context, reliability, and governance as things you design for from the start, not as problems to fix later. It means having one place where the AI's knowledge is managed and kept current, and being able to see exactly what the AI was given, what it produced, and when.
If you are a business leader making decisions about AI right now, the honest truth is that the companies that will win are not necessarily the ones with the most sophisticated models - but those who can deploy AI reliably, catch it when it drifts, and explain it when they have to. That advantage is still up for grabs, but the window will not stay open indefinitely. As regulations tighten and as early adopters prove what is possible, the gap between organizations with a real AI operating foundation and those without one will become very hard to close quickly.
As companies like Promptev Inc work to build that foundation, they are making a bet not on a particular AI model, but on their own ability to use AI responsibly at scale. And that bet is looking smarter every quarter. The key to success in enterprise AI is not just to use a sophisticated AI model, but to have a robust operational backbone that supports it.