Nitin Rakesh is the CEO and Managing Director of Mphasis and coauthor of the award-winning book “Transformation in Times of Crisis.”
On July 20, 1969, the lunar module Eagle was minutes from the moon’s surface when its guidance computer began flashing alarms. A radar left running was feeding it more data than it could process. But it did not crash. It shed its lowest-priority work, protected the tasks that mattered and kept flying.
In Houston, a guidance officer who had rehearsed this exact alarm weeks earlier made the call in seconds: go. A computer with 4 kilobytes of working memory helped land two Apollo 11 astronauts on the moon, not because it was powerful, but because the system around it was disciplined. It knew what to ignore, and a human held the authority to decide.
That distinction between a machine that can act and a system trusted to act well is what enterprise AI must now confront.
All Investment, No Return
Enterprise AI has been defined almost entirely by what it can do, not what it delivers. What can models generate? What can agents execute autonomously? But the question a business leader asks is simpler: What economic impact will this have on the top or bottom line?
Most enterprises cannot answer that. Gartner predicts, “Over 40% of agentic AI projects will be canceled by the end of 2027, due to escalating costs, unclear business value or inadequate risk control.” BCG research shows only about 5% of companies are generating AI value at scale, while nearly 60% report little or no impact to date.
Most enterprises already have the models, the infrastructure and the pilots. What they lack is the mechanism to convert intelligence into governed, accountable, repeatable business action. That mechanism has a name: enterprise agency.
Why Enterprise Agency Matters Now
Enterprise agency is the capacity of an organization to act on its intelligence—to convert what AI knows into decisions, and decisions into measurable outcomes. It is not automation, which makes predefined tasks faster, or autonomy, which lets systems act without continuous human direction. Agency asks a harder question: Is AI helping the business decide better, act responsibly and create real value?
Consider the financial sector. A bank that approves loans faster has improved its efficiency. A bank that improves its credit decisions, reduces risk exposure and expands its responsible lending capacity has created agency. The difference is not the model; it is what the model is asked to do, and whether the organization has the architecture to act on it.
This is the shift ahead: from deploying AI tools to solving business problems with AI.
Outcomes First, Technology Second
If enterprise agency is the goal, AI strategy cannot begin with technology. It must begin with business strategy and intent.
The starting point is a specific outcome: revenue growth, risk reduction, working capital efficiency, customer retention. That outcome shapes the program. It identifies the decision that most directly determines whether the outcome is achieved; the right models and minimum viable data follow.
When AI begins with a tool, organizations retrofit the business case around it. When it begins with an outcome, the technology becomes a means, not the end.
Why Intelligence Without Context Remains Generic
Starting with the outcome exposes a hard truth: Most enterprises lack a single, reliable view of the context behind their decisions.
AI cannot improve a business decision without understanding the reality behind it. That reality is rarely clean data and modern systems. It includes rules, constraints, institutional knowledge and the logic embedded in legacy infrastructure. Enterprise information sits across functions, systems, geographies, workflows, documents and code, plus the experience of people in the business for decades. The same term can carry different meanings in different departments.
Without that context, AI may be powerful. It will not be relevant.
Decision-Ready, Not Just Data-Heavy
Once an enterprise understands the context behind a decision, the question is whether it can improve it before it is made.
For decades, enterprise intelligence has been retrospective. Dashboards told leaders what had already happened. Predictive tools estimated what might happen next. Enterprise agency requires more: the capacity to evaluate choices before action, weighing the likely consequences of each path and selecting the one most aligned to the outcome.
A pricing decision does not improve because AI can act faster. It improves when the enterprise can assess what each option will likely cost and choose accordingly. The question shifts from “What does the data say?” to “What should we do next, and what will it cost us if we are wrong?”
That is the move from hindsight to foresight, where AI moves from analysis to judgment.
Why Governed Action Is The Real Differentiator
The final test is whether that judgment can move into action at enterprise scale, with accountability.
A pilot survives on limited integration, manual work-arounds and enthusiastic sponsorship. A scaled deployment cannot. Enterprise agency requires clear decision rights, auditability, escalation paths and feedback loops that close the cycle, confirming whether an action produced the intended result and improving the next decision.
It is the same architecture that landed Eagle: not raw computing power, but the discipline to shed the right things under pressure, and clear human authority to act.
This is also why agentic AI—systems that act without constant human direction—is necessary but not sufficient. An enterprise full of agents operating without shared context, without foresight and without accountability loops is not more capable. It is more exposed.
Real value emerges when context informs foresight, foresight guides action and action improves the next decision. That closed loop is what converts AI from an experiment into an accountable capability.
The Enterprises That Will Define What’s Next
Enterprise agency will redraw the standards for AI leadership: from effort to effect, from deployment to performance and from experimentation to accountable business value.
As AI matures, access to models and agents will become table stakes. The advantage will belong to enterprises that can convert the intelligence they already have into better decisions, governed actions and measurable outcomes.
Agentic AI may define the next chapter of technology. Enterprise agency will define the next chapter of business.
Forbes Technology Council is an invitation-only community for world-class CIOs, CTOs and technology executives. Do I qualify?