AI not plug-and-play, enterprises need open-heart surgery, says Infosys’ Satish HC
Read Time:3 Minute, 44 Second
Execution challenges, legacy tech stacks and data readiness — not model capability — are the biggest barriers to scaling AI in enterprises, Infosys chief delivery officer tells Moneycontrol
- AI adoption needs deep changes, not just plug-and-play upgrades
- Most AI project effort goes into data prep and system updates
- Infosys shifts to smaller teams and reskilling for AI era
Artificial intelligence (AI) is not a plug-and-play upgrade for enterprises and requires deep changes to core systems and workflows, Infosys chief delivery officer Satish HC has said, cautioning that many organisations underestimate the complexity of adopting the technology at scale.
“Anybody who thinks that AI is plug and play — ‘I will just buy the licenses and plug it in and AI will do the job’ — it’s a disaster,” he told Moneycontrol.
The comments assume importance as it comes at a time when there are growing concerns that AI-driven efficiency could shrink demand for tech services, which has triggered a rout in IT stocks across the world. AI firm Anthropic’s rollout of new Claude plugins has unsettled investors about the demand for traditional IT services, wiping out billions of dollars worth of market capitalisation of IT firms.
Satish described the transition as far more invasive than previous technology shifts. “With AI, we are fundamentally doing open heart surgery to our enterprise. It’s much more complex,” he said.
Satish said the constraint in enterprise AI is not the lack of technological capability but the difficulty of adoption and execution.
“The biggest problem is not with the lack of opportunity with AI. It is the execution problem,” he said. “There is a big gap between the rate of evolution of technology and the rate of evolution of adoption.”
This gap is where IT services firms are positioning themselves, helping enterprises operationalise AI investments.
“They also know they need help in adoption of this, otherwise the amount of money that they’re investing…will not be fruitful. It will not adopt at scale,” he said, referring to technology providers and enterprises.
Productivity gains will come gradually
While generative AI is expected to improve delivery efficiency, Satish said immediate gains from off-the-shelf tools remain limited without enterprise-specific engineering layers.
“Straight out of the box, I can get up to 10 percent productivity,” he said. “To get a higher level of productivity, you have to build frameworks or workbenches that give quality output in the context of a client.”
He described AI adoption as “a journey of foundations”, requiring organisations to encode institutional knowledge so that “humans and machines can talk in the same language”.
Data and legacy modernisation dominate early AI work
Much of the effort in enterprise AI projects still goes into preparing data and modernising systems rather than deploying models.
“In any typical AI project, about 60 percent of the effort goes towards solving the data problems. The model consumption comes after that,” Satish said.
He cited an example where process automation reduced a telecom activation cycle from 65 days to 35 days but further gains required replacing ageing systems.
“If I have to bring it down further, I have to go and transform their core system, which is the system of record, which is very old,” he said.
The shift is also changing Infosys’ internal delivery structure, moving away from scale-driven execution to smaller, specialised teams.
“What was important in the past — process, reliability, scale — is getting transformed into smaller teams, faster throughput, more productivity and more depth,” Satish said.
Infosys plans to continue hiring about 20,000 graduates annually but is introducing differentiated compensation bands ranging from Rs 6 lakh to more than Rs 21 lakh based on skills.
“It is horses for courses,” he said, adding that bridge academies are being used to reskill employees for AI-led roles.
AI will rewrite enterprise stacks, not replace IT services
Satish rejected the view that AI will reduce the relevance of IT services, arguing that each technology wave has required enterprises to redesign systems and operations.
“There have been many tech waves, from PCs to internet to digital to cloud. Every era led to a rewriting of the enterprise stack and reimagining systems and work,” he said. “The same thing is happening with AI, but it’s a much bigger leap.”
He acknowledged that spending remains cautious amid macroeconomic uncertainty, but maintained that the long-term opportunity is significant.
“There is muted spend, there is huge opportunity,” he said.
Happy
0 %
Sad
0 %
Excited
0 %
Sleepy
0 %
Angry
0 %
Surprise
0 %
0
0
Average Rating