Enterprises today have moved past the challenge of whether or not to implement AI and are dealing with how to scale it. Although experimentation has grown quite common within all departments, the shift towards AI for actual outcomes is not yet consistent.
In an interaction with Express Computer, Deepak Pargaonkar, VP – Solution Engineering, Salesforce India, argues that the challenge is less about technology maturity and more about organisational clarity. The difference between success and stagnation, he explains, lies in how enterprises define outcomes, align strategy, and embed AI into core business processes rather than treating it as an isolated initiative.
The real problem is not adoption but direction
At a surface level, it may seem that enterprises are still early in their AI journeys. But Pargaonkar pushes back on that assumption. According to him, adoption is already happening across organisations, just not always in ways that drive measurable value. “Let’s be very, very clear; at an organisational level, adoption has already started. We are using AI in everyday work, whether it is note-taking, creating presentations, or improving productivity. The real question is not adoption but how you translate that into business outcomes like top line, bottom line, productivity, or time to market.”
The gap begins when organisations attempt to move from these everyday use cases to high-impact applications. Too often, AI is approached as a technology experiment rather than a business initiative. “The approach should never be experimentation. You cannot run that as a technology-only initiative. You have to define goals at a business level first. What do you want to improve, and then align technology, data, and talent to that?”
This shift from curiosity-driven pilots to outcome-led execution, is what separates organisations that scale AI from those that remain stuck.
The rise of the agentic enterprise
As enterprises move towards execution, the operating model itself begins to change. Pargaonkar describes this as the emergence of the “agentic enterprise”, where AI agents and humans work together to deliver outcomes, rather than AI functioning as a passive analytical layer. “Agentic enterprise is very simple: humans and agents working together towards a business outcome. The idea is not just to generate insights but to actually execute work.”
This model becomes clearer when applied to real workflows. In sales, for example, AI agents can take over the initial stages of lead qualification, interacting with potential customers, gathering relevant information, and filtering high-intent prospects before handing them over to sales teams.
“What the agents would do is qualify the interest; all that initial interaction is handled by the agent, and once that is done, it is handed over to the salesperson. So now agents and humans are working together, and the number of qualified leads increases significantly.”
Similarly, in customer service, AI can automate routine but critical tasks such as summarising interactions, something that is often skipped due to operational pressure.
“At the end of a customer interaction, AI can summarise the call, the human is focused on the conversation, and the agent ensures that the data is captured properly. That improves both productivity and data quality” he explains, adding that over time, this collaboration extends beyond efficiency into availability.
AI agents enable always-on engagement, allowing organisations to deliver consistent experiences without scaling human effort proportionally. “Agents can provide 24 by 7 support, even for queries, complaints, or actions. That is where the real scale comes in.”
Data is the foundation and not a challenge
If strategy defines intent and agents define execution, data determines whether any of this is even possible. Across enterprises, data readiness continues to be one of the biggest bottlenecks in scaling AI.
Pargaonkar points out that most organisations underestimate the extent of fragmentation in their data environments. “If you don’t have consolidated data, you cannot even answer basic questions, like where your customer data resides. It could be across dozens of systems, and without bringing that together, AI cannot deliver meaningful outcomes.”
The issue is not just about volume but about structure, accessibility, and governance. Enterprises are dealing with a mix of structured and unstructured data, spanning systems, documents, emails, and more, all of which need to be contextualised for AI to work effectively.
“Clean data, consolidated data, is the need of the hour. But that does not mean you have to physically move everything. You can manage distributed data with the right architecture; what matters is how you bring context and control to it,” he says. That is when the discussion shifts from data migration to data strategy, knowing what kind of data we need, where it is located, and how it can be accessed effectively and securely.
Why some industries are moving faster
Although the trend itself is universal and apparent in various industries, not all industries are advancing at an equal pace. Certain industries like finance, healthcare, and consumer-facing companies have made progress more quickly, which can be attributed to the importance and relevancy of use cases. “The thing that really stands out is use cases. In the financial sector, for example, if you can bring down response time from minutes to seconds, you’re seeing the benefit right away. Of course, that would promote adoption.”
The bottom line here is that in a high-throughput setting, AI offers tangible benefits in terms of speed, performance, and efficiency. Organisations in B2C environments, where there are large volumes and multiple interactions, will always see faster success because the impact is visible and scalable. In contrast, industries with fewer transactions or more predictable workflows may take longer to adopt, not due to lack of potential, but because the return on investment is less immediate.
Learning from real-world deployment
One of the most telling indicators of AI’s potential lies in how organisations apply it internally. Salesforce’s “Customer Zero” approach uses its own technology within its operations, which offers a practical perspective on what scaled deployment looks like.
“We have deployed agent capabilities across service, sales, and collaboration, and we are seeing massive business benefits, including significant cost savings and pipeline growth. We have a very tangible story around becoming an agent enterprise, and that helps us guide customers in a more practical way.”
By validating its approach internally, Salesforce is able to move beyond conceptual discussions and demonstrate real-world impact, something enterprises increasingly expect before committing to large-scale AI investments.
From pilots to production: What comes next?
The transition from AI pilots to enterprise-scale deployment marks a critical inflection point. Organisations that succeed are not necessarily those experimenting the most but those executing with clarity and intent.
The path forward requires a combination of strategic alignment, data readiness, and an operating model that integrates AI into everyday workflows through human-agent collaboration.
AI is no longer just a layer of intelligence, but it is becoming a layer of execution.