Hey — It’s Nico.

Welcome to another Failory edition. This issue takes 5 minutes to read.

If you only have one, here are the 5 most important things:

• Hike, a Indian unicorn that started as a competitor to Whatsapp, has shut down — learn why below.

• Six lessons after one year of Agentic AI.

• China tells its tech companies they can’t buy AI chips from Nivida.

• Humanoid robot startup Figure raises $1B+ at $39B valuation.

• OpenAI and Antrophic released data on how users use AI — learn more below.

Today, compliance is your shortest relationship.^AD

We meet. We click. In 15 hours, you’re SOC 2 compliant.
No ghosting, no red flags—just AI agents doing all the grunt work while you close deals.
Our happily-ever-afters:

• Lovable → SOC 2 in 20 hours.

• Bland → $500K ARR in 7 days.

• 11x → $1.2M ARR unlocked.

Fast, easy, no drama. It’s not just done—it’s done in Delve.
Let’s make it official. Book your demo now and get 1k off with code FAILORY1KOFF.

This Week In Startups

🔗 Resources

The mind-virus of Big Venture has become a Pandemic.

Six lessons after one year of Agentic AI.

How to nail your next big talk.

Make LinkedIn your most powerful channel with Playbookz. 20k reach in 28 days or you don’t pay. *sponsored

📰 News

Atlassian acquires DX, a developer productivity platform, for $1B.

Notion launches agents for data analysis and task automation.

China tells its tech companies they can’t buy AI chips from Nivida.

Amazon launches AI agent to help sellers complete tasks and manage their businesses.

💸 Fundraising

Humanoid robot startup Figure raises $1B+ at $39B valuation.

Numeral raises $35M to automate sales tax with AI.

Irregular raises $80 million to secure frontier AI models.

Computer vision startup Ultralytics raises $30M in funding.

Fail(St)ory

From Whatsapp to Gambling

A few weeks ago, India’s real-money gaming industry was hit by a sudden policy shock. Among its casualties was Hike, once celebrated as one of India’s most valuable startups. After thirteen years, hundreds of millions in funding, and multiple pivots, Hike has finally shut down.

What Was Hike:

Back in 2012, Hike launched as a bold challenger to WhatsApp. The idea was to build a messaging app designed for young Indians—complete with stickers, news feeds, games, and even payment features.

For a moment, it looked like Hike might actually pull it off. By 2016, the startup had more than 100 million users, had raised from top-tier investors like SoftBank, Tiger Global, and Tencent, and carried a unicorn valuation of $1.4 billion.

But the global giants didn’t stand still. WhatsApp, Telegram, and Instagram pulled users in with their simplicity, global reach, and relentless product updates. Hike’s super-app ambitions couldn’t keep pace. In 2021, founder Kavin Bharti Mittal shut down Hike Messenger entirely.

That could have been the end of the story. Instead, Mittal and his team pivoted to real-money gaming. Their new product, Rush, let users play casual Indian favorites like carrom and ludo against each other for cash prizes. Within four years, Rush attracted 10 million players and generated over $500 million in gross revenue. For a brief period, it seemed like Hike had found its second life. 

Then came the regulatory hammer. In 2025, the Indian government passed the Promotion and Regulation of Online Gaming Act, banning all real-money gaming. The reasoning was blunt: growing concerns about addiction, financial harm, and even suicides linked to gambling losses. Overnight, Rush’s entire business model was wiped out.

Mittal was candid in his farewell note: the team could have raised more money to try again, but the real question was whether it was worth the climb. His answer—“no.” Not for him, not for the team, and not for the investors who had supported Hike for more than a decade.

The Numbers:

• 📅 Founded in 2012

• 💰 Peaked at a $1.4B valuation (2016).

• 👥 Over 100M users at its peak in messaging; 10M users on Rush

• 🎮 $500M+ in gross revenue from Rush over four years

• 🔥 Shut down in 2025 after India’s blanket ban on real-money gaming

Reasons for Failure: 

• Competing in a winner-take-all market: Messaging apps are famously brutal. The network effects are so strong that only one or two players dominate a region. WhatsApp’s simplicity and global adoption left little oxygen for a regional challenger. Hike could build features, but it couldn’t build the universal network that keeps users locked in.
  

• Pivot fatigue: When Hike shut down its messenger and pivoted to gaming, it proved the team could execute fast. Rush did scale impressively—but it wasn’t the original vision. Real-money gaming was more of a survival play than a long-term conviction. In Mittal’s own words, RMG was “never the destination.” That lack of deep alignment eventually showed.
  

• Regulatory whiplash: The single most direct cause of Hike’s collapse was policy. India’s 2025 gaming ban pulled the rug from under an industry worth $23 billion. Rush, along with other platforms like Dream Sports and MPL, didn’t just face stricter rules—it was banned outright. Startups can live with risk, but uncertainty and sudden regulatory shifts can make a business model unworkable.

• Misjudging market timing: Mittal admits in hindsight that Hike was too early on multiple fronts—first in trying to create a “super app” before Indian internet infrastructure and payments had matured, and later in betting on Web3 gaming without global regulatory clarity. Execution wasn’t the problem. Timing and context were.

Why It Matters: 

• Regulation isn’t just background noise. Hike’s story shows how a startup can be well-funded, well-executed, and still fail overnight if the legal ground shifts.
  

• Survival pivots aren’t the same as conviction bets. Hike managed to reinvent itself once, but the pivot to gaming was never the endgame. When the storm hit, there wasn’t enough belief to justify climbing again.
  

• In winner-take-all markets, “good enough” doesn’t cut it. Hike Messenger was feature-rich and locally beloved, but it wasn’t enough to topple global giants. If you’re building in a market with heavy network effects, you either go global—or get squeezed out.

Trend

How Do People Use AI Anyways?

Last week, both OpenAI and Anthropic released something more interesting than model updates: deep dives into how people actually use their AI tools. The reports are packed with data about who’s using ChatGPT and Claude, what they’re using them for, and how usage is shifting as these products scale.

This is the closest we’ve come to a real x-ray of AI adoption—and the findings say a lot about where startups should be paying attention.

Why It Matters:

• Personal use is winning: ChatGPT is being used less for work and more for everyday life. That means the “Google replacement” angle is turning out to be more real than the “AI assistant at the office” narrative.
  

• Different models, different jobs: Claude is quietly positioning itself as the automation tool for coding-heavy work, while ChatGPT is pulling ahead in consumer-facing tasks like advice, search, and writing.
  

• Inequality is already baked in: Adoption is booming in low- and middle-income countries for ChatGPT, but Claude’s footprint skews toward wealthier regions, setting up a possible productivity gap.

What GPT Users Do

OpenAI’s research shows just how much ChatGPT has become a part of daily life. In June 2024, about half of all conversations were work-related. By mid-2025, that share dropped to just over a quarter—even as total usage exploded from 451 million daily messages to 2.6 billion.

So what are people doing with it?

The top buckets are “asking” for guidance (nearly half of all messages), “doing” tasks like writing or summarizing (about 40%), and “expressing” opinions or feelings (the smallest slice, around 11%).

In practice, that means most conversations are about practical help: tutoring, explanations, recipes, advice on purchases, or clarifying news and events. Writing tasks—emails, documents, blog posts—still play a big role, but they’re dwarfed by the “everyday helper” use case.

The demographic shifts are equally telling. ChatGPT’s early user base was overwhelmingly male (around 80%), but now it’s closer to parity. Nearly half of all usage comes from people under 26, making it a distinctly youth-driven phenomenon. And while adoption is strong in wealthier countries, OpenAI’s data shows growth is 4x faster in low- and middle-income regions—a pattern that looks very different from Claude.

For entrepreneurs, this suggests something interesting: ChatGPT is becoming more like an infrastructure tool for everyday decision-making, not just a productivity hack. It’s the internet’s new front door.

What Claude Users Do

Claude’s usage paints a very different picture. Anthropic splits activity into two categories: automation (AI handles the task with little back-and-forth) and augmentation (AI and human work together). For most of 2024, augmentation was slightly ahead. But in early 2025, automation overtook it for the first time—nearly 50% of all tasks are now full delegation.

The most common directive? Coding. Claude users lean heavily on it to write, debug, and automate software tasks. Anthropic’s transcript analysis shows a clear rise in directive-style prompts where users say “do this” and trust the result without iterating. This hints at growing confidence in AI output—at least in specialized technical contexts.

Geographically, Claude’s footprint is concentrated in high-income regions. The US accounts for the largest share, with California dominating in absolute usage, while Washington, DC has the highest per-capita adoption. Outside the US, Israel is a standout—its per-capita usage is about seven times higher than± expected.

Anthropic’s own warning is worth noting: adoption patterns mirror past technological divides, like electricity and the internet, where richer countries got a head start and captured more of the gains. With Claude already skewing toward automation and technical use, this inequality could compound fast.

What It All Means

Taken together, the two reports show how usage patterns aren’t just about “AI in general” but about the roles specific products are carving out.

• ChatGPT looks like it’s becoming the casual AI for the world—used by students, workers, and hobbyists for everyday problem-solving.

• Claude, on the other hand, is sliding into the professional niche where precision and automation matter most.

For startups, the lesson is simple: don’t assume “AI adoption” means the same thing across platforms. If you’re building for consumer scale, the ChatGPT ecosystem is where your users already are. If your product leans on automation, coding, or IT infrastructure, Claude’s user base might be the better fit.

And one last insight: the shift from “augmentation” to “automation” should make every founder think carefully. Delegation is rising because users are starting to trust AI with full tasks. That changes the value chain. If AI is no longer just a collaborator but a silent worker, what does that mean for your product, your customers, and the kinds of services they’ll pay for?

Help Me Improve Failory

That's all of this edition.

Cheers,

Nico