Monday , Sept. 30, 2024, 1 p.m.
News thumbnail
Technology / Thu, 25 Apr 2024 Moneycontrol

How this TCS executive-cum-food influencer uses Gen AI to manage close to a million followers

In his food influencer avatar MasalaLab, Ashok posts informative and incisive short videos on culinary science, history and anthropology and has nearly 1 million followers across platforms. Rather, he deploy his own custom-made generative AI assistant. Story continues below Advertisement Remove Ad"I manage close to a million subscribers on social media with no team and no studio. For the AI assistant, he uses a premium version of ChatGPT, the chatbot developed by US-based OpenAI , and APIs or application programming interfaces. It is tuned to give me a 250-word script," Ashok said.

TCS' Global Head of consulting for AI.Cloud unit Krish Ashok. He is also known as a content creator who produces short informative videos on food science, history and so on | Photo: Moneycontrol

As the global head for Tata Consultancy Services' (TCS) AI.Cloud Advisory and Consulting, Krish Ashok wears multiple hats. But in the world of social media, it is the figurative toque that he dons he is better known for. In his food influencer avatar MasalaLab, Ashok posts informative and incisive short videos on culinary science, history and anthropology and has nearly 1 million followers across platforms.

However, this techie, unlike his many other influencer peers, does not take the help of any agency to map out his content creation journey. Rather, he deploy his own custom-made generative AI assistant.

Story continues below Advertisement Remove Ad

"I manage close to a million subscribers on social media with no team and no studio. My focus is on good storytelling and everything else is minimalist and automated. For that, I've created my own GPT assistant to help me with scriptwriting," Ashok said on the sidelines of Google Cloud Next event held earlier this month in Las Vegas.

He has 800,000 followers on Instagram, 123,000 on YouTube and 89,000 on X (formerly Twitter).

For the AI assistant, he uses a premium version of ChatGPT, the chatbot developed by US-based OpenAI , and APIs or application programming interfaces. "It has been trained on my book, my columns, my information, my tone of voice and it gives me a draft of the script that I want to record... It is tuned to give me a 250-word script," Ashok said.

He additionally uses Perplexity's AI model to fact-check the script to prevent any "hallucinations". AI hallucinations are incorrect or misleading results that AI models generate.

Earlier, in the "pre-Gen AI era", as he terms it, Ashok used to focus on taking notes efficiently. For that, smartphones, note-taking tools and cloud were some of the technologies he used to fall back on for his creative process.

"I've always been on the lookout for this kind of automation. A lot of the problem with creativity is staring at a blank slate. If I have something that's even 50 percent of the way there, then I'm very productive. Then I can just edit it, I can quickly tweak and then I can tell the story that I want to tell," he added.

Story continues below Advertisement Remove Ad

In 2023, Ashok released his maiden book, "Masala Lab: The Science of Indian Cooking". His next book is due to be released this year, he said.

He admits that time has been an issue ever since he took on his current role at TCS in October 2023. Around that time, TCS integrated all of its cloud units—Microsoft, AWS—in a single horizontal. Ashok heads the consulting unit for that horizontal.

So what does the work entail? Essentially, as Ashok said, "For instance, if someone wants to build a knowledge assistant for their unit, we do everything from going and finding out the right problem to solve, to prioritising in terms of business value, risk. And then look at the value chain and see where the maximum impact is. Then we have built a platform that allows us to rapidly do a POC (proof of concept) very quickly."

The turnaround rate for such POC for standard use cases, Ashok said, has come down to a few weeks, adding that complicated cases, though, often take months.

However, Ashok admits that the conversion from POCs to production is very hard. "This technology is very expensive. Once you scale it to 20,000, the token costs and everything else becomes very expensive," he said.

logo

Stay informed with the latest news and updates from around India and the world.We bring you credible news, captivating stories, and valuable insights every day

©All Rights Reserved.