Rahul Jha, VP – Cloud, Gen AI & Cybersecurity, Visionet Systems
Enterprises are no longer experimenting with GenAI. Most CIOs today are sitting on a long list of use cases, each with different complexity and different ROI expectations. But scaling GenAI is not about making one use case successful. And enterprise ROI will not come from a single implementation, it will come from successfully executing multiple use cases at scale. That’s where most organizations will fail.
The problem is not building AI. The problem is readiness.
The Myth of Easy AI Scale
There is a growing belief that once pilots work, scaling will follow. In reality, most enterprises are not ready for scale because they lack a strong AI or intelligence foundation. This foundation starts with data readiness, or what is better understood as knowledge readiness. In many cases, 30 to 40 percent of the effort still goes into preparing data before anything meaningful can begin. Simultaneously, building agents is no longer the challenge. That is becoming easier by the day. What remains hard is deploying them securely, consistently, and at scale.
Meanwhile, there is a tendency to treat AI readiness as an extension of cloud modernization. There is definitely some overlap, but this is a broader shift. This is intelligence modernization, arguably even business modernization. Cloud platforms are becoming the gateway, and enterprises will rely on a mix of providers, models, and frameworks. But no organization can afford to be locked into one ecosystem. What is required is a flexible harness which can work across models, frameworks, and providers, with everything consumed through APIs and governed by strong security controls.
Security Will Be the Breaking Point
If there is one area where enterprises are most unprepared, it is security. Everything in cybersecurity today is built for a deterministic world where systems behave predictably. GenAI does not. The interface is now conversational; the behavior is dynamic. The same input can lead to different outcomes. Agents can reason, act, and interact with other systems. This fundamentally changes the threat landscape. Hence, security is no longer just about protecting systems, it is about controlling behavior.
Moreover, enterprises now have to deal with entirely new attack surfaces, from prompt injection to unpredictable outputs, from agent interactions to autonomous decision-making. And this is not a single-layer problem. The lifecycle from idea to product has shrunk from months to days, and that entire pipeline needs to be secured. Consumption introduces its own risks. Managing one agent is different from managing thousands. Also, data security remains critical. Governance and compliance requirements are evolving. All of this has to be addressed simultaneously.
One of the clearest examples of this shift is agent identity. Agents are not human, but they can act with authority. Managing their authentication, authorization, and control is still an unsolved problem, and it becomes even more complex at scale. Hence, enterprises will need to track how agents are created, how they interact, and when they should be retired, something existing systems were not designed to handle.
At the same time, AI is becoming a core part of security itself. It is already used across operations, compliance, testing, and monitoring, making AI-driven security a baseline capability. However, securing AI systems is still evolving. This gap will ultimately determine how successfully enterprises can scale.
What needs to change
Enterprises need to shift their focus from building AI to preparing for it. It will only start with investing in knowledge readiness. Equally important is establishing a strong operational backbone: clear governance for agents, robust identity and access controls, and consistent frameworks that can work across models and platforms without lock-in. Most importantly, organizations must rethink execution. Scaling GenAI is not about deploying more tools or launching more pilots. It is about building systems that can run continuously, securely, and reliably at scale.
In the end, the enterprises that succeed will not be the ones that move fastest, but the ones that prepare best.