In a clear signal to India's burgeoning AI sector, the latest cohort of the Google and Accel Atoms accelerator has pointedly avoided funding so-called 'wrapper' startups. From over 4,000 applications, the program selected just five companies, each focused on building substantive AI technology rather than superficial features layered on existing models.

"About 70% of applications were wrappers—adding a chatbot to old software without reimagining the workflow," said Accel partner Prayank Swaroop. Many other rejected ideas crowded well-trodden areas like marketing automation, offering little new ground.

The chosen startups, which will receive up to $2 million from Accel and Google's AI Futures Fund plus significant cloud credits, reflect a push for deeper integration. They include K-Dense, an AI 'co-scientist' for research; Dodge.ai, building autonomous agents for ERP systems; Persistence Labs, focusing on voice AI for call centers; Zingroll, a platform for AI-generated film; and Level Plane, applying AI to industrial automation.

Jonathan Silber of Google's AI Futures Fund noted the selections align with expectations for meaningful AI adoption. The program doesn't mandate Google's models, seeking instead real-world feedback. "If a company uses an alternative model, that means Google has work to do," Silber said, describing a feedback loop where startup insights directly inform model development at Google DeepMind.

The application surge—nearly four times previous cohorts—was dominated by enterprise software ideas, with 75% focused on productivity or developer tools. Swaroop expressed a desire to see more ambition in sectors like healthcare and education, suggesting where the next wave of Indian AI innovation might emerge.

Source: TechCrunch