Looking back at 2025: how our AI led software startups evolved

India's deep tech landscape is witnessing a quiet but meaningful evolution in artificial intelligence-led software, with private startups demonstrating that AI adoption need not be revolutionary to be valuable.

Rather than claiming to transform entire industries overnight, a new generation of entrepreneurs is building practical tools that address specific operational challenges across verticals—from data automation to customer on-boarding, privacy protection to cross-border payments.

Our own portfolio of investments in AI-led software startups illustrates the range of opportunities Indian founders are pursuing. Among our AI-focused investments, startups are tackling diverse problems: Airboxr automates data analysis for e-commerce businesses, helping Shopify merchants make faster decisions without extensive manual reporting.

Looppanel uses AI to streamline qualitative user research, providing automated transcription and thematic analysis that cuts research timelines significantly. Wingman, acquired by US-based Clari in 2022, built conversation intelligence software that analysed sales calls to surface insights—a category that has since matured globally.

The spectrum extends into specialized domains. Protecto, which raised $4 million in seed funding in late 2023, addresses data privacy concerns in AI deployments by betokening sensitive information while preserving data utility for model training. Rocketlane focuses on customer on-boarding for B2B SaaS companies, a niche that emerged as critical during the pandemic when remote implementations became standard.

Skydo, which secured RBI approval as a cross-border payment aggregator in September 2025, serves over 12,000 Indian exporters by simplifying international payments with transparent pricing and zero forex markup.

Threading through these ventures is a pragmatic approach: identify friction points in existing workflows and deploy AI where it demonstrably improves efficiency. Threado built AI assistants for internal knowledge management, helping customer support teams access information across tools and documents instantly. TurboML, meanwhile, is investing $12 million to develop AI foundational models proficient in Indian languages, addressing the country's linguistic diversity — a challenge that global models handle inadequately.

Government backing has provided crucial infrastructure. The IndiaAI Mission, approved with a Rs. 10,372 crore outlay for five years, has expanded computing capacity from under 1,000 GPUs two years ago to over 38,000 by October 2025, offered at subsidized rates of Rs. 65 per hour. This expands access for startups that previously found compute costs prohibitive. The mission's seven pillars — encompassing compute infrastructure, dataset platforms, foundation models, skills development, startup financing, and responsible AI governance — reflect a comprehensive approach to ecosystem building.

Public-private collaboration has taken tangible form. The IndiaAI Startups Global programme, launched in March 2025 in partnership with Station F and HEC Paris, selected ten Indian AI startups for international acceleration, providing European market access. Four startups—Sarvam AI, Soket AI, Gnani AI, and Gan AI—were chosen to develop India's sovereign large language models under government support. Such initiatives signal policy continuity beyond budgetary announcements.

India's AI startup ecosystem raised $990 million cumulatively by mid-2025, a 30 percent year-on-year increase, though this remains modest compared to global counterparts. The country's AI market is projected to reach $17 billion by 2030, with current revenues approaching $280 billion across the broader tech sector.

What distinguishes India's trajectory is less about absolute scale and more about deployment patterns — 87 percent of enterprises now actively use AI solutions, with adoption concentrated in BFSI, retail, healthcare, and industrial sectors.

Challenges persist. Access to specialized AI research talent remains constrained, with India holding just 4 percent of the global AI talent pool at researcher level. Regulatory compliance, particularly around data localization under the Digital Personal Data Protection Act, adds complexity for startups building cross-border solutions. Funding, while growing, requires startups to demonstrate clear unit economics rather than rely on growth narratives alone.

The Indian AI-led software story for 2025 is one of incremental progress across multiple fronts — founders building viable businesses in established and emerging niches, government providing enabling infrastructure, and enterprises gradually adopting tools that prove their value. This measured approach may lack the drama of breakthrough claims, but it builds a foundation for sustained relevance in the global AI economy.