Accel India Accelerator Selects 5 Promising Startups Without ‘AI Wrappers’

Artificial Intelligence Startup Ideas: A Closer Look

Numerous artificial intelligence startup concepts still rely heavily on existing models, functioning as mere “wrappers” on top of them. However, as these AI model creators introduce more features, investors are cautious of startups that could easily become redundant.

During the evaluation of over 4,000 applications for the joint AI accelerator program for Indian startups led by Google and Accel, a prevalence of “wrapper” ideas was observed. Despite this, none of these ideas made it to the latest cohort, as revealed by Accel partner Prayank Swaroop in an interview with TechCrunch.

In November, Google and Accel announced the launch of the AI-focused Atoms program, aimed at supporting early-stage startups developing AI products associated with India. Selected startups in the current cohort stand to receive up to $2 million in funding from Accel and Google’s AI Futures Fund, along with up to $350,000 in cloud and AI compute credits from Google.

A significant portion of the rejected applications were categorized as “wrappers,” startups that added AI features like chatbots onto existing software without innovating new workflows using AI, according to Swaroop.

Many of the remaining denied applications fell within competitive sectors such as marketing automation and AI recruitment tools, where originality was lacking. Swaroop noted the challenge for startups in these areas to differentiate themselves.

Interestingly, this year’s program saw a substantial increase in applications compared to previous Atoms cohorts, with a significant number of first-time founders applying.

The growing AI ecosystem in India predominantly focuses on enterprise applications, which was reflected in the submitted applications. Around 62% of the proposals centered around productivity tools, while another 13% targeted software development and coding. This indicates that the majority of the applications revolved around enterprise software rather than consumer products.

Jonathan Silber, co-founder and director of Google’s AI Futures Fund, highlighted that the five selected startups closely align with areas where Google anticipates deeper real-world adoption of AI.

The program does not mandate startups to exclusively use Google’s models. Silber mentioned that many companies combine multiple models based on their workflow requirements. The primary objective is to gather feedback from startups on the performance of Google’s models in practical applications.

Insights gleaned from these startups can then be utilized by Google DeepMind teams to enhance future models, creating a symbiotic relationship between startup experimentation and AI development. Silber emphasized the importance of refining Google’s models based on real-world usage.

The selected startups for this year are:

• K-Dense: Developing an AI “co-scientist” to expedite research in areas like life sciences and chemistry;
• Dodge.ai: Creating autonomous agents for enterprise ERP systems;
• Persistence Labs: Focusing on voice AI for call center operations;
• Zingroll: Building a platform for AI-generated films and shows;
• Level Plane: Applying AI to industrial automation in automotive and aerospace manufacturing.