India AI Impact Summit — Reflections

Walking into the India AI Impact Summit felt less like an industry conference and more like a live wiring diagram of the country’s AI ambitions. In one afternoon, you could hear about drone racing societies from Universities, an AI-powered doctor at Hospitals, a bot helping farmers understand crop patterns, and a startup selling multilingual AI agents to small businesses (kirana shops, as they call it here in India) all in the same breath.

What struck me most wasn’t any single product or platform. It was the sheer density of activity and the unmistakable sense that India isn’t waiting to catch up its building its own lane. Here are the highlights, the surprises, and the questions worth chewing on.

A Quick Snapshot from Our Summit

50+Companies & Platforms Met
10+Opportunities Identified
5State / Government AI Pilots
4Partners Met

The 5 Big Themes from the Summit

01.

Agentic AI Is No Longer a Buzzword — It’s Shipping

Nearly every conversation touched on agents. Not chatbots. Not copilots. Agents’ autonomous systems that plan, reason, and execute. HCL’s AI Force 2.0, Accenture’s agentic AI course, Evvolv.ai’s AI workers for SMEs, and Airo’s AI recruitment platform were all built around this paradigm. The shift from ‘AI that answers’ to ‘AI that does’ is very real, very messy, and very exciting.

“Money is not short. What we need is companies to energize innovation.”

— Aramco Grand Challenges representative, on why they are running open industrial AI challenges globally

02.

India-First Language Models Are Having a Moment

Sarvam AI was everywhere literally and figuratively. With 105B+ parameters, support for Indian languages, a conversational agent reaching 4 crore (i.e. 40 million) people, and 24% WhatsApp engagement rates, they are the most credible homegrown LLM story right now. Gnani.ai, Vyrah (13 regional languages), and Jio’s own AI stack are not far behind. This is not a niche play it’s a structural bet on the next 500 million internet users.

03.

Healthcare AI Is Being Built in India, for India

Eka.care (used by Apollo), Aarogyam AI (on Gemini models, integrated with ABHA ID), Datalet Healthcare, Neurodx (brain scans), and Microsoft’s M Health 360 and Asha Bot the healthcare corridor at the summit was arguably the most mature. ABHA integration, real-time transcription, multilingual patient communication, and diagnostic imaging were all live products, not prototypes.

04.

The Infrastructure Layer Is Getting Crowded

GPU-as-a-service is officially a race. Cyfuture, HCL, Neysa, and Velocis were all competing here. The interesting meta-observation: inference is increasingly cheap enough that you don’t need to own GPUs at all. The value is moving to what runs on top.

05.

Government as Customer and Co-Builder

The State of Assam declared itself India’s first ‘AI-first state.’ The State of Tripura deployed Kyma AI for school students. ONDC’s MyStore Mitra is using Google’s infrastructure. Jio Krishii is building for farmers. Bhashini and Sarvam are powering a government-backed AI teacher. The public sector is not just regulating AI, it’s procuring it aggressively.


Engineering & Product Signals

What’s worth paying attention to from an engineering and product perspective:

Governance & Observability

ServiceNow’s App Engine + AI asset governance layer is mature. Their ‘white space’ model (identifying what’s unaddressed in a client’s stack) is a sharp GTM framing.

Multimodal Design Tools

Adobe Firefly is a diffusion model (not LLM), commercially safe, trained on licensed data. Layered output for color extraction, generative upscaling, and harmonization are genuinely useful. Photoshop on web with agent-like layer creation is live.

Voice AI Maturity

ElevenLabs is live with 100+ languages, 11 including Indian ones. 5,000 hrs/month volume commitments. AI dining and music generation are adjacent use cases.

Vertical AI Stacks

Sarvam’s Arya OS for AI and Jio’s full-stack (JioKrishii, JioArogya, JioCreator, Jio Intelligence) suggest vertical OS-level plays are coming for India.

Robotics & Physical AI

Simplismart, Addverb (human-robot collaboration), Neurodx, and Soket are all in early-but-real territory. Worth tracking.

Under-the-Radar Bets

Google is actively building quantum computing. DeepMind’s WeatherNext predicts cyclones. Meta Advantage+ is maturing for campaign automation.

Radar Ping

Odoo is being positioned as a Zoho alternative. Zoho itself has 55 applications, strong on CRM + helpdesk. Worth benchmarking both.


Narrative opportunities, competitive signals, and tools worth exploring

The “Why” is The New UX

Mastercard’s ‘Reasoning Intelligence’ framing explaining why a recommendation is made is a compelling positioning angle we can borrow. Transparency in AI decisions is becoming a differentiator, not just a compliance checkbox.

AI Is Eating Our Toolkit

Meta Advantage+ is automating campaign management. LinkedIn’s Hiring Assistant shortlisted top 50 from 13,000 profiles. Virsa.ai scrapes LinkedIn to build persona-targeted email + WhatsApp sequences. Our audience is going to encounter all of this. We need a point of view.

Pricing Signals from the Market

Evvolv.ai charges ₹10K/user/month for 1,000 tasks. Meritto (enrollment CRM) starts at ₹18K–30K/year. These are the benchmarks SMB and mid-market buyers will use to evaluate us.

Copyright-Clean AI Is a Selling Point

Firefly’s ‘commercially safe, not trained on public data’ message is resonating. Clients are increasingly asking about data provenance. This is a content opportunity.

Partnership leads worth following up

Brandworks is working on government and strategic initiatives. Nirmaan.org runs underprivileged hiring programs. These could be meaningful CSR-adjacent or partnership plays.

Vernacular Is Not Optional Anymore

Sarvam, Gnani.ai, Vyrah, Jio all pitching regional language AI. If we’re not thinking about vernacular content strategy, we’re missing a growing segment.

Quick win

Microsoft’s AI Skills Navigator is a good benchmark for how to package upskilling narratives for B2B audiences.


Warm leads, sectors with budget, and people to follow up with

Hot Lead

Salesforce’s startup program explicitly looking for partners and startups to plug into their ecosystem.

Meritto

Meritto (enrollment CRM, 1,200 clients, ₹18K–30K/year) is actively demoing. Could be a channel partner or integration opportunity.

Evvolv.ai

Evvolv.ai sells AI workers to SMEs, Haldirams is a client. B2B sales motion similar to ours, could be a reseller or co-sell candidate.

EY Cybersecurity

EY’s Cosmos cybersecurity platform + Cyber AI CoE. Vulnerability management + compensatory controls (MFA, firewall). Enterprise budget, clear pain point.

Zing HR

Zing HR AI-enabled HR, positioning as Keka alternative. SMB HR tech is a growing wedge.

Adani Connex

Adani Connex infrastructure scale, exploring AI and connectivity plays.

Education Sector

Education: Kyma AI (₹500/student, Tripura government contract), Bharat1.ai, Gemini’s Sarathi guided learning. State government procurement is an underrated channel.

Watch the space

Fairvaluation.com is building AI valuation tools for investors + startups. Could be useful for our own fundraising narrative or as a partner for deal flow.

The One Thing That Stayed With Me

At a panel discussion, someone said: ‘System design is where AI still lacks.’ It was the most honest thing said all day. Every company in that room had a demo. Many had customers. But the ones that will win are the ones that don’t just deploy AI they redesign the system around it. That’s the harder, slower, more important work. And frankly, it’s the conversation most companies haven’t had yet.
That’s the opening we should be walking through.