2025: When AI Tests Indian IT
Indian IT is stepping into 2025 with stakes as high as can be, as its multi-billion-dollar bets on generative AI face the ultimate challenge of delivering tangible results. [See below the top six predictions for Indian IT next year]
Despite deploying hundreds of AI solutions in 2024, the revenue growth has been below expectations. “Indian IT is shifting from generic AI models to tailored solutions for clients,” said Mohandas Pai, former CFO of Infosys, in an exclusive interview with AIM. The question now is whether companies like TCS, Infosys, Wipro, HCLTech, and LTIMindtree can scale these efforts into transformative, profitable ventures in the year ahead.
Automation and AI adoption are also reshaping the workforce. While leaner and agile teams may emerge due to productivity gains, AI-specific skills are in demand. In a marked shift in hiring priorities, firms like TCS plan to onboard 40,000 freshers this year, focusing on AI expertise.
“IT services might start with a 2:1 workforce reduction as AI boosts productivity,” noted Vaibhav Domkundwar, CEO of Better Capital. Upskilling has become non-negotiable as companies prepare their employees for this transformation.
Indian IT’s next frontier lies in creating revolutionary AI products, but the path is steep. Building large-scale models like ChatGPT requires resources that are still out of reach. Yet, industry leaders remain optimistic. “This phase is about perfecting services and preparing for the big leap,” said former NASSCOM VP Viswanathan K S.
As 2025 unfolds, the industry must prove whether its AI ambitions can redefine its global relevance or if the true revolution is still to come.
Best Firm for Data Scientists >>
Best Firm for Data Scientists is one of India's biggest workplace certification platforms in data science.
In a first, Hansa Cequity has been certified as one of the Best Firms for Data Scientists for the third consecutive year, while IDFC First Bank and Aays have earned this recognition for the second year.
To nominate your organisation for the certification, fill out the form here.
Top Stories of the Week >>
Llama 3.3 Just Made Synthetic Data Generation Effortless
Meta recently released Llama 3.3, a 70B parameter model designed to revolutionise synthetic data generation. Combining the performance of its 405B predecessor with drastically lower costs, it supports multilingual output in languages like Hindi, Thai, and Portuguese, empowering developers to create tailored, privacy-compliant datasets. While Meta gears up for Llama 4, promising stronger reasoning and multi-modality, Llama 3.3 sets the stage for accessible, domain-specific AI innovation worldwide. Read more here.
The Year English Became a Global Programming Language
In 2024, AI made English the gateway to programming, democratising a field once limited to experts fluent in languages like Python or Java. Tools like GitHub Copilot and Claude now let anyone describe what they need in natural language and receive working code, empowering non-developers to create software. But this shift isn’t just automation—it’s artistry. Mastering prompt engineering is key to unlocking AI’s potential. With barriers falling, the question isn’t whether you can code but what you’ll create next. Explore the revolution.
Fine-Tuning is Dead, Long Live Reinforcement Fine-Tuning
OpenAI recently unveiled reinforcement fine-tuning (RFT), a technique that makes traditional fine-tuning obsolete. Unlike conventional methods, RFT teaches models to reason, and not just repeat. This breakthrough lets organisations fine-tune AI for high-stakes domains like healthcare and law with minimal data—sometimes as few as 12 examples. Read more here.
Why is No One Talking About Adversarial Machine Learning?
While AI advancements dazzle the world, a silent threat—adversarial machine learning (AML)—casts a long shadow. From misclassifying stop signs to compromising credit models, attackers subtly manipulate inputs to derail AI systems by exploiting vulnerabilities. These sophisticated attacks include data poisoning, evasion tactics, and model theft, threatening industries like finance, healthcare, and autonomous driving.
The solution? Robust defences: adversarial training, regular audits, and explainable AI. As AI’s role deepens, embedding security at every stage of its lifecycle isn’t optional—it’s essential. Learn more here.
AIM Videos >>
GenAI is Going to Permeate Every Aspect of Society
In this compelling episode of the Simulated Reality podcast by AIM, Phil Le-Brun, enterprise strategist at AWS, touched upon the transformative powers of generative AI. Drawing comparisons to the printing press and electricity, Le-Brun emphasised that we are merely scratching the surface of AI's potential at this point.
AIM Events >>
MLDS 2025: Can You Afford to Miss This?
Packed with keynotes, hands-on workshops, live coding, and unmatched networking opportunities, MLDS 2025 is where the future of tech is shaped.
The conference will bring together 2,000 developers, 100 expert speakers, and 600 organisations to redefine the future of AI, generative AI, cloud computing, and more. Seats are limited—secure your pass now, or risk being left behind!
AIM Nuggets >>