Mohammed Rafee Tarafdar, chief technology officer (CTO), Infosys, says that in recent times, several small language models for India-specific needs have been launched and with the growing developer base, “we should see a lot more GenAI applications implemented for Indian and global markets.” In his opinion, as GenAI is becoming a general-purpose technology, it will eventually be applied to all the sectors and industries. In this interview, he speaks to Sudhir Chowdhary on some of the key trends in AI adoption and Infosys’ gameplan to stay ahead in a competitive and rapidly changing market. Excerpts:
What are some of the top trends set to define industries in 2025?
In 2025, I see a lot of AI initiatives currently under rollout, to be scaled across enterprises, and businesses will start realising some measurable business value along the lines of cost, growth, better experience, and risk protection. We are seeing increased investments in scaling inferencing which improve the reasoning capabilities, thereby enabling the agentic systems to be used to eliminate tasks and re-engineer the processes.
We will move from operational processes to core industry specific business processes like customer onboarding, credit decisioning, risk management etc. As the small language models become more specialised and are able to deliver higher accuracy at lower cost, the adoption of these models in enterprises is likely to accelerate.
Aren’t firms struggling to find ROI for AI projects?
We are seeing some of our clients already realise value in their AI projects. The ROI is coming in productivity, efficiency, and cost, but as more and more re-engineered business processes are rolled out by enterprises, I see value being realised for better customer net promoter score (NPS) and higher growth opportunities.
What is Infosys’ larger plan in terms of staying ahead in this competitive market?
Infosys continues to invest in four areas relevant to enterprise AI – Data readiness & Model engineering, Agents & Agentic Platforms, Responsible AI, and AI-Ready Talent. We recently launched our own small language models developed for banking and IT Ops.
We now have a base enterprise model using which we can create industry specific models easily. To help our clients re-engineer the business processes, we are developing AI agents and agentic platforms that can be used to automate and implement these processes. We continue to invest in our responsible AI capabilities, to ensure we implement all AI solutions using our 12-factor responsible by design framework.
Which industry verticals are fast adopters of GenAI?
As Gen AI is becoming a general-purpose technology, it will eventually be applied to all the sectors and industries. However, today we are seeing large-scale adoption in the financial services, telco, manufacturing, CPG, logistics, energy and services industry.
What new steps will ensure effective deployment of AI?
Organisations will have to set up a Responsible AI office to ensure that the organisational policies and processes are evolved to cover the entire AI engineering lifecycle. In addition, technical & legal guardrails will have to be implemented to ensure that privacy, bias, security issues, and identified and mitigated upfront.
Most organisations may not have the talent for implementing Responsible AI practices, and hence will have to invest in hiring the existing talent to build deeper competency . At Infosys, to address these challenges, we have developed a 12-factor framework that provides solutions for explainability, reproducibility, fairness & bias, safety, privacy, compliance, IP protection, sustainability and governance framework.
