Although 43% of enterprises claim high agentic AI expertise, nearly half of them allocate less than 20% of their tech budgets to it, according to Deloitte’s State of AI report.
At first glance, this disconnect between confidence and commitment seems too much to ignore. However, Ashvin Vellody, partner and chief strategy & innovation officer – consulting, Deloitte Asia Pacific, says the situation is more nuanced.
“Every technology has a normal adoption cycle,” Vellody explained in a conversation with AIM.
“What’s different about agentic AI is that the interest and intent to adopt it is one of the highest ever seen at this stage in any technology’s life cycle.” But that doesn’t mean enterprises are rushing in headfirst.
In what Vellody calls a “reverse adoption,” agentic AI tools like ChatGPT first gained traction among individual users before enterprises caught on. “That’s where the evolution of the ‘prosumer’—professional and consumer—comes in,” he said.
People started using these tools for everyday tasks, and now they expect their companies to offer the same capabilities. Yet for enterprises, it’s not just about plugging in a tool. These are brownfield systems with years of legacy data, infrastructure, and processes. “You can’t just flip a switch,” Vellody said.
But it’s not as bad either. According to Deloitte’s latest Tech Trends – India Perspective 2025 report, Indian businesses are increasingly adopting small, multimodal models to meet demands for faster, more efficient, and targeted solutions, with AI leading the transformation across sectors.
In April, Deloitte announced its largest investment yet with Google Cloud by introducing over 100 ready-to-deploy AI agents built on Agentspace and Gemini. These agents aim to transform customer and employee experiences across industries.
The company is also co-developing the Agent2Agent protocol with Google and ServiceNow, which will enable interoperability among AI agents across platforms and cloud providers.
During the announcement, Jason Salzei, chair and CEO of Deloitte Consulting LLP, said, “Clients are getting flooded with information about agents… With collaborations like Google Cloud and ServiceNow, we aim to be a one-stop shop.”
Deloitte claims a productivity boost of over 30% in areas like contract redlining, credit loan automation, and technical data transformation.
It’s Not Popcorn
The buzz around agentic AI is that it’s on par with the internet revolution, at least that is what the makers, or the first movers of this generation, claim. This holds weight, but Vellody believes that though the technology could prove more disruptive than anything we’ve seen before, the real focus should be on its outcomes—boosting productivity, increasing efficiency, and reinventing business models.
That said, meaningful change doesn’t happen instantly. “It’s not instant like popcorn,” he said, underscoring the need to address ethics, privacy, and hallucination risks. In some cases, slowing down is the smartest way to speed up.
In India, Vellody sees enterprises falling into three categories: early adopters, co-creators, and wait-and-watchers. “Roughly 15–20% are aggressive early adopters,” he estimated. “But all three approaches are okay—it depends on company culture and leadership vision.”
Deloitte’s State of AI report highlights that around 70% enterprises feel excited about GenAI, but 27% also express uncertainty. This explains why Indian enterprises are also moving slowly.
But innovation isn’t absent; it could be better. “Can Indian enterprises push more on innovation? Yes. Are they already doing it? Also yes,” he said. “Even companies in traditional sectors—like tile or switchgear manufacturing—are exploring agentic AI within their context.”
Enterprises aren’t resisting agentic AI, they’re navigating it. Cautiously, strategically, and with an eye on long-term value. As Vellody put it: “It’s not about being first. It’s about being right.”
Why Consulting Firms Are Winning AI Revenue
With recent earnings from Accenture and ServiceNow highlighting strong AI contributions, and even OpenAI hiring consultants, the question arises: Are services and consultant firms making more from AI than startups?
Read: Accenture’s $1.5 Billion AI Wake-Up Call for Indian IT
“The pattern is simple,” Vellody said. “When a new technology comes up—whether it’s a car without a steering wheel or agentic AI—you need someone to show customers how to use it. The people building the tech can’t scale that instruction, so they partner with consulting firms.”
Startups may dominate headlines, but enterprise spending remains deliberate. The headlines focus on big tech and private equity bets, not the practical brownfield adoption journey that most enterprises are navigating.
At Deloitte, agentic AI deployment is structured into three stages: defining where to play (strategy), how to win (execution), and how to measure (impact). “We see success in manufacturing, healthcare, and more,” Vellody said. “But ethics and governance are non-negotiable.”
No enterprise wants an AI model to hallucinate. This is where consulting firms like Deloitte, EY, and others come into play. Even EY India is heavily focused on AI, with a $1.4 billion investment in AI over the next five years.
Similarly, Deloitte is investing $3 billion in generative AI by 2030 and is increasing its AI investments by 30% in FY26. It is clear that consultant firms have grabbed AI by its long tail.
Vellody’s interest in AI is not new, though. He began his career as a software developer in the late ’90s, and admitted he’s dabbled again thanks to agentic tools like Lovable and Devin. But he cautioned against alarmism about coding jobs disappearing.
“These tools make good developers better,” he said. “But if you don’t know what to do at all, they’re of limited use. You still need context and strong prompt engineering.”
He said that one of the most important things that Deloitte, and he personally, want to make sure of is that these technologies are not taken out of context, either by accident or by design. In Vellody’s current role, which he calls “a privilege,” he gets to play in the zero-to-one space—building from concept to MVP.
“If I weren’t doing this,” he mused, “I’d still be chasing use cases in this field. There’s just so much potential.”
The post ‘Agentic AI isn’t Instant Like Popcorn’ appeared first on Analytics India Magazine.
