Salesforce says Indian SMBs are moving faster on AI than large enterprises, but costs and data gaps remain

Indian small and medium businesses (SMBs) are moving faster than large enterprises in deploying artificial intelligence (AI), as simpler decision-making and lower implementation barriers allow them to test and scale new tools more quickly, according to Salesforce. However, the pace of adoption does not necessarily translate into business-wide integration. High technology costs, fragmented data, security concerns and the difficulty of demonstrating returns remain significant hurdles for businesses looking to move beyond experimentation.

Salesforce’s Small and Medium Business Trends Report found that 78% of Indian SMBs surveyed were either using or experimenting with AI. Among businesses using AI, 95% reported more efficient operations, while 97% said the technology had improved productivity, revenue or customer experience. The survey covered more than 200 Indian SMBs.

Arun Kumar Parameswaran, Executive Vice President and Managing Director, Sales and Distribution, South Asia, at Salesforce, said smaller businesses were often able to move faster because they had fewer layers of decision-making and less legacy technology to contend with. “The fastest deployments we see for AI are in SMBs. For the simple reason that you don’t have legacy,” he said, in a conversation with Fortune India. “You have one man who holds the purse strings. At the end of the day, he’s making the decision.”

Smaller businesses also tend to start with simpler use cases, making it easier to assess whether an AI deployment is delivering value. “The use cases are fairly simple. Easy to deliver. Easy to conceive business value. You can very quickly decide if you’re getting value or not getting value,” Parameswaran said. Large enterprises, by contrast, must align multiple departments on the use case, technology architecture, security requirements and expected returns before deployment. That process can slow implementation even when budgets are available.

AI adoption is rising, but the business case remains uneven

Salesforce’s survey points to growing AI use among Indian SMBs, but Parameswaran said the wider market remains at different stages of digital maturity. He estimated that 5-10% of businesses are digitally mature, while roughly 40% are transitioning from digital transformation to AI. The bottom 50%, he said, are still largely operating through spreadsheets and WhatsApp.

The World Economic Forum (WEF), in its 2025 report, Transforming Small Businesses: An AI Playbook for India’s MSMEs, estimates that AI could unlock $490 billion to $685 billion in economic value for the country’s MSMEs by 2030.

Parameswaran said enterprises were under pressure to demonstrate AI initiatives, sometimes leading to pilots that were not tied to a clear business outcome. “I don’t know that I would call it an adoption. I would say 47% have piloted,” he said, referring to a figure discussed during the interview. He attributed the proliferation of pilots to boards asking executives what they were doing with AI, prompting some companies to pursue projects without a sufficiently defined business need.

For him, successful AI deployment rests on three factors. “AI is about three critical things that need to be handled and answered. First is trust. Second is context, and third is outcome and value.” He said a lack of context could make AI inaccurate and expensive, while weak governance could undermine employee and customer confidence. The cost of deploying AI also needs to be assessed across the entire system, rather than through software licences alone. “Total cost of ownership is everything. It’s the cost of the platform, it’s the cost of the people that are building it, the cost of people that will have to run it and everything associated with it,” he said.

The economics of automation can also differ sharply between India and markets where labour is more expensive. Parameswaran recalled a US customer saying that deploying Salesforce’s Agentforce had reduced the cost per contact-centre call from $7 to $2. An Indian customer, he said, responded that the cost of a call at its own contact centre was just ₹7. “If I go to that customer whose cost of a call in a contact centre is 7 rupees and I say, ‘Pay me $2 for doing this with AI,’ why would they?” he said.

From productivity to revenue, AI’s value is shifting towards outcomes

Parameswaran said Indian businesses were increasingly looking to AI to drive revenue and improve customer experience, rather than focusing solely on cutting costs. “The shift in this part of the world is definitely more towards driving top line and experience compared to efficiency and productivity,” he said. “No CFO is going to say no to revenue.”

Parameswaran also pointed to growing interest in outcome-based pricing, in which customers pay for a defined business result rather than the volume of tokens consumed by an AI model. AI voice applications are another area gaining traction in India, particularly as support for regional languages improves. Parameswaran said outbound voice applications were already widespread, while inbound voice use cases still had challenges to overcome before they could see comparable enterprise adoption. He also flagged the difficulty of ensuring consistent results from generative AI models.

Parameswaran cited a Salesforce customer whose AI model was deprecated a day before deployment. After the customer upgraded to a newer version, 50% of the outcomes changed. “The problems didn’t change, the data didn’t change, but the outcomes changed overnight from one model to another,” he said. For regulated sectors such as financial services, consistency is particularly important. “If you’re the RBI, for example, you will expect that 10 out of 10 times, the answer has to be the same,” he said.

For Indian businesses, the challenge is moving beyond experimentation to AI systems that can demonstrate measurable business value while meeting requirements for cost, security and reliability.