India’s rapidly expanding artificial intelligence startups is attracting thousands of founders, but investors are becoming increasingly selective about the ideas they support. As AI technology evolves, venture firms are paying closer attention to startups that build original workflows and deep AI capabilities, rather than simply layering basic features on existing models.
During the evaluation of more than 4,000 applications for the joint AI accelerator for Indian startups run by Google and venture firm Accel, many of the proposals fell into a similar pattern. According to Accel partner Prayank Swaroop, a significant portion of these applications focused on building so-called “wrapper” startups—companies that place simple AI features on top of existing platforms. However, none of those ideas made it into the latest cohort of five selected startups.
The AI-focused Atoms program, announced in November by Google and Accel, aims to support early-stage startups developing AI products connected to India’s ecosystem. The startups chosen for the new cohort will receive up to $2 million in funding from Accel and Google’s AI Futures Fund. In addition, they will gain access to up to $350,000 in cloud and AI compute credits from Google to accelerate development.
Swaroop explained that nearly 70% of the rejected applications were “wrappers.” These startups often added AI-powered features such as chatbots to existing software but lacked deeper innovation.
“Wrapper” startups layered AI features such as chatbots on top of existing software but “were not reimagining new workflows using AI,” Swaroop said.
Meanwhile, many of the remaining rejected applications focused on highly saturated sectors, including marketing automation and AI recruitment tools. According to Swaroop, startups in these segments frequently struggle to differentiate themselves because multiple companies are already competing with similar solutions.
The large number of similar ideas was partly driven by the growing interest in AI entrepreneurship in India. This year’s program received almost four times more applications compared to previous cohorts of Accel’s Atoms initiative. Notably, a significant share of applicants were first-time founders, highlighting the enthusiasm surrounding AI-driven businesses.
At the same time, India’s AI startup landscape remains heavily enterprise-focused. Swaroop noted that about 62% of the submissions centered on productivity tools, while another 13% targeted software development and coding solutions. As a result, nearly three-quarters of all applications focused on enterprise software, rather than consumer-facing AI products.
Interestingly, Swaroop also pointed out that he expected to see more ideas related to healthcare and education, sectors where AI could potentially create meaningful social impact.
From Google’s perspective, the final selection reflects areas where the company expects AI adoption to deepen in real-world applications.
Jonathan Silber, co-founder and director of Google’s AI Futures Fund, said the startups chosen for the program closely match industries where AI can drive practical and scalable impact.
Importantly, the program does not require participating startups to rely solely on Google’s AI models. Silber emphasized that many companies use multiple AI models depending on their workflows, which allows them to optimize performance and flexibility.
The initiative also serves a broader strategic purpose for Google. By observing how startups use different AI technologies in real-world scenarios, the company can gather valuable insights that help improve future models.
Participating startups share insights with Google DeepMind teams, enabling engineers to refine AI capabilities based on real product usage.
Silber described this approach as creating a feedback loop between startups and AI developers.
“If a company is using an alternative model, that means Google has work to do to build the best model in the market,” he said.
With growing interest from founders and investors alike, the country’s AI ecosystem is likely to evolve toward more specialized, scalable, and transformative solutions.
