Hey {{first_name | there}},

India has the highest generative AI adoption rate on the planet. 73% of its population already uses generative AI. 92% of Indian workers use AI tools several times a week. 

We consume AI like chai - constantly, enthusiastically, and without thinking twice.

Now open any global LLM leaderboard. 

Go ahead. Scroll through it.

You'll find GPT-5.5, Claude 4.8 Opus, Gemini 3.5 Pro, Qwen, DeepSeek, Mistral, Minimax…

Keep scrolling.

Keep going.

You won't find an Indian model in the top 20. Not even close.

Think about that for a second. The country where nearly everyone uses AI can't build one that anyone outside India chooses to use.

When Sam Altman visited India in 2023 and said Indian startups couldn't build a ChatGPT rival with a few million dollars, the country went into overdrive. 

Krutrim launched. Sarvam launched. BharatGen launched. Everyone clapped.

But here's the uncomfortable truth that nobody in the press wanted to say out loud:

Sarvam AI, India's most talked-about and government-backed LLM, marketed itself primarily on being Indian, not on being good.

The pitch was: "We understand Indic languages. We support 22 Indian dialects. We are proudly Indian."

Nobody asked the benchmarks question. Because the benchmarks were… not flattering.

According to the Artificial Analysis benchmark data, global frontier models from OpenAI, Anthropic, Google, and Alibaba cluster around intelligence scores of 57–60, with context windows nearing 1 million tokens. 

Indian models aren't in the same conversation. 

Nationalism is not a benchmark. Period.

Remember Krutrim? India's first AI unicorn. 

I know you probably have forgotten about it. Even I did before I started writing it.

In January 2024, Krutrim raised $50 million at a $1 billion valuation. It announced LLMs, AI assistants, enterprise tools, cloud infrastructure, and its own AI chp: a full-stack AI ecosystem all at once.

By March 2026, the company had gone from 550+ employees to roughly 150–160. 

Its consumer chatbot, Kruti, is dead. Its semiconductor division has been effectively dismantled. Its LLM work has been paused. 

And it's now pivoting to… AI cloud infrastructure services. Basically, renting compute to other companies.

What happened? It tried to do everything, burning $4 - 5 million a month. With no clear product moat. 

It's not just a startup failure. It's a mirror for how India thinks about deep tech: announce big, raise fast, stretch thin, quietly collapse.

Let's not be naive and blame one company. This is structural.

1. We don't fund R&D. Full stop.

India spends 0.65% of its GDP on research and development. South Korea spends 4.5%. The US spends nearly 4%. China spends 3%. You cannot build a frontier AI model on hope and nationalism. 

You need compute, researchers, and years of patient capital.

The government's India AI Mission has allocated about ₹10,372 crore (~$1.2 billion),  a serious number on paper. 

But compare that to what OpenAI, Anthropic, and Google spend in a single quarter, and you start to understand the gap.

2. Our investors want a US template first.

A founder on the StartUpIndia subreddit said it perfectly: 

"I pitched a genuinely differentiated B2B SaaS idea to three Mumbai-based angel investors. 

All three asked the same question within minutes - 'who in the US is already doing this?' When I said no one, they lost interest immediately. 

Not because the idea was bad. Because they didn't have a Western template to validate it against. 

That's not investing. That's photocopying."

3. The IT services trap.

For 30 years, India's economic identity was built on being the world's cheap execution engine, IT services, BPOs, back-office ops. TCS, Infosys, Wipro, in 40 years, none of them built a single product the world remembers. We exported brains and imported nothing back in terms of product DNA.

Now AI is doing exactly what those services do, cheaper, faster, without needing a visa. The $250 billion IT industry isn't just disrupted. It's being deleted in slow motion, and Infosys is buying AI tools to do it themselves.

4. No safety net = no risk-taking.

In India, one failed startup can sink a family for years. There's no social safety net. So the smartest people optimize for stable salaries, not moonshots. IIT students choose finance over frontier research. PhDs leave for the US because DRDO pays interns ₹5,000 a month. ISRO internships are literally unpaid.

"China saw AI coming and threw state money, talent, and urgency at it. India formed a committee." - This line from Reddit has 1000+ upvotes for a reason.

Here's what I think:

You're probably not going to build India's DeepSeek. Neither am I. 

But you don't have to.

The opportunity isn't in building the model. 

It's in building on top of the models that exist, for problems that are deeply Indian, deeply specific, and deeply underserved. Voice agents in regional languages. 

AI tools for India's 63 million SMBs that still run on WhatsApp. 

Agentic workflows for sectors like agriculture, education, and healthcare where global AI simply doesn't understand the context.

But to do any of this well, you need to be around people who are actually building, not just talking about building.

The conversation in most Indian circles is still stuck at "which LLM should I try?" 

The real builders are already asking, "Which agent architecture handles this use case, and how do I make it production-ready in INR, not USD?"

If you want to be in those conversations, and I mean real ones, not LinkedIn fluff, come join the AI community I'm building.

- Aashish

P.S. If you're an Indian founder currently building something AI-native, reply and tell me what you're working on. I'm genuinely curious and I might feature it in the next edition.

Last month, 48% of visitors to documentation sites across Mintlify were AI agents, not humans.

Claude Code, Cursor, and other coding agents are becoming the actual customers reading your docs. And they read everything.

This changes what good documentation means. Humans skim and forgive gaps. Agents methodically check every endpoint, read every guide, and compare you against alternatives with zero fatigue.

Your docs aren't just helping users anymore. They're your product's first interview with the machines deciding whether to recommend you.

That means: clear schema markup so agents can parse your content, real benchmarks instead of marketing fluff, open endpoints agents can actually test, and honest comparisons that emphasize strengths without hype.

Mintlify powers documentation for over 20,000 companies, reaching 100M+ people every year. We just raised a $45M Series B led by @a16z and @SalesforceVC to build the knowledge layer for the agent era.

Make Your Docs Agent-Ready