Taking early-stage AI startups from idea to impact at TechSparks 2023

At TechSparks 2023, Ram Ganesan, CEO of Sivi AI, and Ajith Sahasranamam, Founder and CEO of Ongil, dove into the GenAI landscape in India and recounted their experiences building for the segment.

Taking early-stage AI startups from idea to impact at TechSparks 2023

Sunday October 22, 2023,

3 min Read

With artificial intelligence (AI) startups becoming the talk of the town in 2023, early-stage startup founders came together at TechSparks 2023 in Bengaluru to talk about life as an early-stage innovator in AI.

Ram Ganesan, CEO of Sivi, and Ajith Sahasranamam, Founder and CEO of OngIL, dove into the Generative AI (GenAI) landscape in India and recounted their experiences building for the segment.

“When starting out, you don’t usually have access to the right kind of data for your product. Without the right model, what do you sell? Building the data flywheel is the hardest challenge during the initial days,” said Ajith.

Ajith’s startup Ongil helps senior management make data-driven decisions even when they have limited access to data, analytics support, and time. Ongil is an instant messaging platform for senior management across industries and functions that speeds up analytics tasks ranging from data collection to insights.

“One approach for an early-stage startup is to rely on publicly-available data sets to train their models. The second approach is to generate your own data using samples from public sources,” Ajith explained, adding: “Then, the big challenge is building a minimum viable product where users are okay with 85-90% accuracy.”

Besides relevant datasets and models, early-stage GenAI startups also require computing power and infrastructure, such as cloud technology, to run and scale their products. While India doesn’t have a significant presence in any of these segments, the country’s strength lies in its vast talent pool and entrepreneurial energy.

And, it can leverage this zeal for entrepreneurship to build more AI startups that solve specific, narrow use cases instead of broad ones like multi-format text, image, and video creation. Working on narrower problem statements requires smaller quantities of datasets, computing power, and infrastructure, according to Ram.

“There are many non-general models, which startups can build for specialised use cases. So they don’t really need the large compute infra typically required for large language models,” he said. “Startups can operate with smaller datasets. They just need to make sure their business fundamentals and value proposition are solid.”

Sivi, a design-centric GenAI company, has built its specialised models and dataset to generate business visuals (marketing creatives, email headers, product banners, etc.).

The two founders also spoke about the investment landscape for AI startups in India—which, according to a report by NASSCOM—saw Indian GenAI startups raise over $475 million in funding between 2021 and May 2023.

“Traditional startup success stories around funding have usually revolved around the evaluation of metrics relevant to industries like ecommerce and SaaS. But each sector is different. In AI, everybody is building new tech, and it’s not easy to figure out new metrics and take a bet on tech,” Ajith said.

Edited by Suman Singh