The project, which is at a very initial stage right now, is looking to harness the vast internal code, data and resources that the country’s largest software services firm has. The solution will be built through in-house algorithms that use large language modelling functions similar to generative artificial intelligence (AI) chatbots such as ChatGPT but for enterprise code generation.
“The way we look at it, it (generative AI) uses past code, data and experience to learn,” Subramaniam told ET. “And over the many years that TCS has been in business, I can use all of my knowledge as a base. So, if that technology uses and generates code that I have taught the algorithm using TCS proprietary data, then the outcome is something that I am willing to license.”
He said TCS is working on use cases to build proprietary generative AI algorithms to build code and the process has to mature further. He believes enterprise usage of these generative AI solutions still requires finetuning in terms of ownership and IP of the end products, hence enterprise solutions will see demand.
Also read | TCS payroll generates Form 16 documents in record time
Subramaniam, who retires in May next year, expects the Mumbai-based software firm to roll out some of these solutions before the end of his tenure. “The focus is really to build models and using those models, generate the code that can be deployed,” he said.
Discover the stories of your interest
He said if TCS wants to provide secure and contextual generative AI solutions to accelerate client projects, it cannot depend only on multiple large language tools from different providers but should have its own solution.“Everybody (clients) is ready to use generative AI. But then, as a supplier, I am obliged to deliver them something which is free of any infringement, and full ownership of the intellectual property,” Subramaniam said.
Also, the industry will have to wait for regulations governing these solutions to mature further to realise the full potential of the technology, he said.
Prompt Engineers
Subramaniam sees a huge opportunity in creating new roles for “prompt engineers” within the organisation who can effectively communicate contextual queries to large language models for enterprise requirements.
“So, one of the things that we are hearing in the market is that the way you prompt the questions to generative AI platforms is very important, right? So, we are conceptualising a role called ‘prompt engineers’ within the organisation,” he said. “We are yet to get approval on it, but we know that this role will be as critical as having engineering, mathematical or analytical skills.”
TCS sees an opportunity to integrate its generative AI offering with its low-code software development platform MasterCraft.
This will be similar to Copilot – a solution to auto-generate code, launched by Microsoft-owned GitHub and OpenAI in October 2021, to help write software programs faster.
“As of today, over 100 clients are using the Mastercraft tool,” Subramaniam said. “We have not put ChatGPT-like solutions for data pattern intelligence into it. But when we do eventually, our vision for MasterCraft to be a (intelligent) low-code platform will come true.”
He said TCS has already built a number of internal use cases using generative AI solutions while also working with global partners.
