MyShell Shows That AI Startups Are Really Just Issuing Tokens

"Crypto Hype Solves the AI Dilemma"

On this account, I've repeatedly argued that star AI projects are essentially Web 3 at their core.

The past two days, I finally met the real deal.

There's an actual AI + Web 3 product being built seriously — MyShell.

Compared to it, products like Cluely and Headai look positively naive. The hype gap isn't even close. When a professional Web3 team enters AI, it's dimensional warfare.

I suddenly realized: the faith-driven model of AI entrepreneurship has been perfectly solved by Web3's self-circulating hype machine.

You've probably seen MyShell's PR pieces already. One says it wants to be the Douyin of the AI era, another says it wants to be AI's Instagram. Basically, they're pushing an AI content community story (taking notes from YouWare, eh? 🤓)

Enough with the bravado. Open MyShell's website (app.myshell.ai) and one glance tells you what they're actually doing.

It's just a grab bag of AI toys: a chat window on the left, a pile of small gadgets on the right. The popular ones include Lucky Kisses, which generates short videos of female celebrities blowing you kisses; Labubu Maker, which converts your uploaded images into Labubu-style versions...

Built for social media — whatever's trending, they make it.

MyShell's recently promoted Agent feature is even less interesting.

Painfully familiar interface: a chat window, you type text, MyShell generates a corresponding Agent.

I don't know if Web 3 teams just lack engineering chops. This interaction pattern is so mature, so utterly uncreative — and MyShell still can't get it right.

I tried three tasks, all failures: a simulated Google Chrome dino game that just spun endlessly with no progress; a "Fat Cat" video generator where the frontend was built but couldn't call the MiniMax API key I entered, so no video; finally I lost patience and asked for a Gomoku game, which turned out to require manually entering coordinates to make moves.

On pure product capability, MyShell doesn't even match YouWare. At least the latter can actually produce a bunch of small toys.

If MyShell were a straightforward AI product, I could already call it: this thing belongs to the category of hype products that lack even original narrative ability. A stitched-together Frankenstein of Manus invite codes, YouWare community storytelling, and a pile of Meitu-style mini-games — call it AI Frankenstein.

But here's where MyShell actually gets interesting: it's an AI + Web 3 product.

It simply can't be bothered to engage at the product level. Click the bottom-left button on the homepage and you switch from Classic Mode to Web 3 Mode — this is its real battlefield.

So what's MyShell's Web 3 story?

Creator token economy. Any little toy you make can be minted into a token.

The platform launched the Shell token in February. When users interact with your Agent, the platform settles Shell tokens to you.

Total Shell supply: 1 billion. Currently about 290 million in circulation. Price: $0.163. Top developers on the platform can cash out roughly $1,000 at a time.

This "creator economy" narrative has been beaten to death in crypto circles.

But MyShell's brilliance lies in perfectly stitching together the two bubbles of AI and Web3, solving AI entrepreneurs' biggest pain point: the inability to make their story self-consistent.

Take a Vibe Coding community like YouWare, which insists on telling a story about "frontend mini-toys as the next content format." But here's the problem: defining what a product is represents an extremely scarce capability. How could ordinary users' hastily cobbled-together toys possibly compare to 4399's mini-games? This narrative has a fatal flaw — no exit path.

Web3 solves everything at once.

Web3's core is consensus. Completely detached from reality, it's the internet industry's Protestant Reformation — shifting from performance-driven to justification by faith.

Could MyShell's toys be any more worthless than NFTs? Could ShellAgent be any more vaporous than Dogecoin?

Doesn't matter. Just believe. MyShell tells a logically self-consistent closed-loop story with AI + Web3: if you're willing to believe it has value, then the Shell token really has value. You say users' cobbled-together toys are useless, but users can issue tokens with them — as long as someone believes, they're useful.

And MyShell's strongest suit? Operations. Battle-tested operational capability forged in Web 3, an industry of pure hype merchants.

ShellAgent is running an invite-code beta; the system shows over 500,000 people in queue (thanks to Guancha for the invite code — I've written plenty of short takes there too). The Discord group has 160,000 members. Official accounts get tens of thousands of views per tweet on Twitter. And a swarm of crypto and AI KOLs constantly repost MyShell-generated Labubu videos.

AI entrepreneurs should seriously study MyShell. Applying Web 3 operational tactics to AI products is dimensional warfare.

Here's my quick summary of MyShell's operational playbook:

Hype two years ahead: Started points airdrops in 2023, requiring users to complete daily tasks — "follow official Twitter, repost designated tweets, join Discord" — to earn Shell points, convertible to Shell tokens later. (Though in the end, 600 points exchanged for 1 token...)

Crypto hype flywheel: Place PR across major exchanges like Binance and crypto media to attract token users. Meanwhile, MyShell's one-click token issuance design incentivizes users to promote on Twitter, driving traffic to their own tokens.

Twitter hype matrix: English main account, Chinese community account, core team personal accounts, platform top creators, exchange accounts, various KOL accounts — multi-language, high-intensity mutual reposting and @-mentions, GOING TO THE MOON 🚀🌖 As they say, the master leads you through the door; the rest is up to you.

MyShell's hype methods look simple, explainable in a few sentences. But they've been running airdrops for two years, using crypto's battle-tested token issuance playbook to compete with AI products — a listed company beating a startup team, dimensional warfare in the hype arena.

I love you crypto people so much. Crypto leads the internet by at least five years, having already developed a high-level self-circulating hype ecosystem.

While honest products like YouWare spend real dollars on community incentives, MyShell just issues its own Shell tokens — 1 billion total supply, 700 million unvested, issue however they want.

From an AI product perspective, MyShell runs entirely on faith, pure virtual economy.

But — any Web 3 folks want to chime in? Over in your world, does MyShell count as the real economy?

(Images in this article generated by ChatGPT o3, with Gemini 2.5 Pro assisting in writing. Made a video version 👇🏻)