Infosys has developed 4 small language fashions for banking, IT operations, cybersecurity and broader enterprise use. These small language fashions, with 2.5 billion parameters, leverage a few of the firm’s proprietary datasets, in line with Salil Parekh, CEO and MD of Infosys.
Creating Small Language Fashions
“We’re growing over 100 new generative AI brokers for deployment inside our purchasers. We’re working carefully with the generative AI accomplice ecosystem to develop joint options for our purchasers, a number of of them on the platforms of the companions,” Parekh added in the course of the Q3 2025 earnings name on January 16, 2025.
Additionally Learn: IBM and L’Oreal to Construct AI Mannequin for Creation of Sustainable Cosmetics
Actual-World Purposes of Generative AI
Explaining some examples of the work Infosys is doing for purchasers within the generative AI house, the CEO mentioned the corporate has developed “a generative AI-powered analysis agent that generated complete options inside seconds for requests made for the product assist groups of a giant know-how firm.”
And for an audit company, Infosys has “created three audit brokers to intelligently automate a number of duties for an expert providers firm,” Parekh mentioned, including, “With Infosys Topaz, our generative AI-powered providers and options, we’re deepening our enterprise AI capabilities.”
Strengthening Enterprise AI Capabilities
“We proceed to strengthen our enterprise AI capabilities, significantly specializing in generative AI, which is witnessing growing shopper traction,” mentioned Salil Parekh within the earnings launch.
“Whereas the main target stays on value optimisation, spending in the direction of new progress areas like AI, cloud adoption, cyber safety information and analytics is noticed,” the corporate mentioned.
Additionally Learn: Infosys AI Bulletins: 2024 Yr in Overview
Classification of Nature of Work
Responding to a query on the classification of the character of labor (on AI), Parekh mentioned, “So in the present day, AI is one thing the place many consumers are doing completely different, completely different applications. So it isn’t like the normal tech, which had that form of a view and when industries have been getting again, the discretionary was growing and in any other case it was extra value. So in the present day, we see the spend is broad based mostly. The top final result generally could possibly be the associated fee for their very own progress, nevertheless it’s not like that simply put into a type of buckets in the present day not less than. Because it turns into extra mainstream, we’ll be capable to see how they use it. As we speak, there’s a broader utilization of AI inside corporations that is occurring.”
Infosys CEO on AI Washing
Individually, in the course of the firm’s Q3 FY25 convention name with media, the CEO, responding to a query on AI Washing, mentioned, “On generative AI, I believe you talked about the AI washing. So, I am not conscious of that inside Infosys, however you could concentrate on that exterior with another corporations. We’re very clear on what we’re doing with Generative AI.”
He continued explaining, “These small language fashions, simply for instance. As we speak, we’ve a number of discussions with purchasers the place they want to use the small language modes that we’ve constructed. So, how are they constructed? They’re constructed through the use of the proprietary information that we’ve, for instance on banking or IT operations, after which use some very commonplace Trade, or on this case, horizontal information after which the shopper builds their very own into that small language mannequin.”
Actual Generative AI Work
“Some purchasers are asking us to construct a small language mannequin of their very own. So, for instance, with a telco shopper, they wish to construct their own-let’s say, firm X Telco-their personal small language mannequin, which we’re serving to them with as a result of we’ve the platform for it. That is actual generative AI work what we’re doing. You then take a look at brokers. To present you some examples, we’ve constructed, for a client-this is precise work, not only a proof of concept-where we’ve constructed a analysis agent for a shopper, a big tech firm, the place they’re now utilizing that of their product space to assist how queries are checked out, and the place their very own folks and their very own clients can use this agent.”
“We’ve constructed an agent for audit work for an expert providers firm – the three completely different agents- that are actually serving to that firm to extra effectively and with fewer errors do what they’re doing of their audit exercise. So, the work we’re doing in Generative AI, we really feel, is main within the trade. We’re very clear in the way it’s getting used throughout as a result of these are actual initiatives with purchasers.”
Head Rely
Throughout the name, he additionally famous that the corporate’s headcount grew by over 5,000 sequentially to now over 323,000 staff worldwide.
