Gen AI PoCs will be scaling to production incrementally, won’t go from 10% adoption to 100% immediately, says TCS CTO Harrick Vin

(L-R) Siva Ganesan, Head, AI.Cloud at TCS and Harrick Vin, Chief Technology Officer, TCS

Tata Consultancy Services’ (TCS) Chief Technology officer Dr Harrick Vin said that companies are incrementally scaling from generative artificial intelligence (AI) proof of concepts (PoCs) to production and are unlikely to quickly grow from 10 percent scaling to almost 100 percent immediately.

While companies are doing PoCs, where the focus is more on value and less on risks, putting it into production requires a different threshold for what is acceptable and that’s the journey most companies are embarking on now in 2024, Vin said.

“It’s a slow path to production for sure…Initially, when a new technology comes up, everybody wants to play with the new shiny toy which happened in the last one year. The focus is now shifting from potential to value. After that it will be scaling value,” he said.

Vin was speaking to Moneycontrol on the sidelines of the launch of TCS AI WisdomNext platform launch on June 7.

He shared that as companies move to scaling up their generative AI PoCs, a technical challenge will be to ensure that the work that’s being moved into production is indeed effective. “It has to be consistent, cost effective and responsible,” he said.

The second big challenge, he highlighted, will be to scale value in a changed management, as the role of people would change in an organisation with AI adoption, and companies will need to retrain people continuously. That will take some time to internalise and implement.

“Expecting magic overnight is probably highly overestimating the potential in the short run,” he said.

Vin added, “The adoption will keep getting better incrementally. Whether it will go from 10% to 100%, I don’t think so. It will keep on incrementally growing.”

TCS AI WisdomNext

High costs of deploying generative AI and lack of regulatory framework around data security, as highlighted by Vin, have been the two biggest challenges for companies in the recent times, which have been a hindrance in scaling up generative AI projects.

To solve these, TCS has launched TCS AI WisdomNext, a platform that aggregates multiple generative AI models including Vertex AI, AzureOpenAI, OpenAI, WatsonX, AWS, and OpenSource in a single interface. It also comes with a plethora of industry blue prints of solutions and use cases to choose from.

So just like an e-commerce website, companies could pick the solution or use case they are looking for, then explore the generative AI models available in a drop down, compare costs and effectiveness of the solutions offered and check out to see the final pricing accordingly.

This will help companies select right large language models (LLMs) at optimised costs and experiment with multiple models on the same platform. Additionally, there will be options for customising solutions too as many of the customers have very unique problem statements.

TCS has already showed it to over 100 existing customers, said Siva Ganesan, Head, AI.Cloud Unit, TCS.

Ganesan told Moneycontrol, “This will also involve a huge consultative role that we are playing as the customers wants guidance on which model to pick, which are the use cases suited to it. Actually for some problems you don’t need generative AI at all. Traditional AI models would do the job. The advice TCS gives to customers is on what choice of generative AI, whether it should be a combination of Gen AI and traditionally AI.”

He added, “We are doing a lot of investments in this space. We have done a lot of skilling, training and certification initiatives too. Earlier in the cloud platforms, and now in AI and Gen AI space.”

Till date, TCS has trained over 300,000 employees in AI skills. CEO and MD K Krithivasan had earlier called it one of the largest AI-ready workforces in the world.

TCS’ AI.Cloud business unit was launched last year saw its deal pipeline doubling to $900 million worth of AI and generative AI projects as of Q4FY24. TCS has won over 200 engagements in AI so far this year.

Krishna Mohan, deputy head, Al.Cloud, TCS said that of these 200 projects, a significant number of them have already used WisdomNext platform in some form.

Ganesan added that while almost all segments have benefitted but Banking, financial services and insurance (BFSI), retail, consumer goods, travel, life sciences and healthcare, manufacturing have looked into different industry blue prints available on the platform the most.

Also read: AI will result in higher productivity and job losses; need to reskill people to prepare them: N Chandrasekaran