Every app wants you inside their app. They want you to say how lovely their new £100,000 dashboard is. The strokes and shadows and spacing, it’s just right. My productivity is going to fly now, thank you.
There’s ego in asking someone to reorganise their life around your interface. Maybe it’s better, but there’s ego in it. A company looking at your day and thinking: we should be the centre of it.
Last year, everyone in Silicon Valley decided to build a browser. Collectively there was a race. A mass-overreaction that the browser layer winner would win the data moat forever. The Browser Company. Perplexity. OpenAI. They were all internally chanting Peter Thiel’s thesis at each other - that you need to be 10x better than the incumbent (Google Chrome, Microsoft Edge) to get people to switch.
But the equation was wrong.
They were benchmarking against Chrome. And maybe they were 10x better. But the competition wasn’t Chrome. The competition was Chrome with a ChatGPT tab open. Chrome with Perplexity fetching you the answer and bringing it straight back to the chat. That was the bar. 10x better than that. None of them cleared it.
Notably OpenAI shipped Atlas and Perplexity shipped Comet, and there were others. WIRED’s reviewer disabled the sidebar after a few days, preferring “the fullscreen version of the internet.” I personally don’t know anyone using Atlas, or comet. You might? I bet you don’t.
The reason for this browser war was that the browser was the ultimate surface-area grab. It’s one big attempt to win the data moat. To own the infrastructure underneath. You win against the SAAS company’s agents, if your agent is the layer below. It has higher clearance of access, and by extension becomes a better product. That’s the general rule and thesis and actually I don’t really disagree. But.
The best case study against this model, the antithesis, over and over again, is Anthropic.
Everyone rushed to make an IDE. Cursor raised at $29 billion. Remember the Windsurf Saga? OpenAI tried to buy Windsurf for $3 billion. Google poached Windsurf’s CEO for $2.4 billion and built Antigravity. Cognition grabbed what was left for $250 million. Six billion dollars of musical chairs over who gets to own the dashboard developers stare at.
Then Anthropic released Claude Code. In the terminal. No GUI. No IDE. No dashboard. Interoperable, simple, plug and play. It didn’t care if you were in Cursor, Windsurf, Antigravity, VS Code. Now Claude Code is the most-used AI coding tool on earth. Zero to number one in eight months.
Everyone rushed to make a browser. Anthropic came out with a Chrome extension that lived inside the browser you already use. With your cookies and logins. There was no migration. One reviewer called it “the most thoughtfully designed browser agent in 2026” specifically because it met users where they already were.
Anthropic built MCP, open-sourced it, made it model-agnostic, and donated it to the Linux Foundation. Now there’s 5,800+ servers, 97 million monthly SDK downloads, and it’s adopted by OpenAI, Google, Microsoft.
Maybe you see the pattern emerge. Everyone asks people to move. Anthropic meets them where they are. Everyone tries to own the surface. Anthropic doesn’t and wins.
Following that thread, there’s also OpenClaw. Not Anthropic. Solo dev Peter Steinberger. You hatch an open-source AI agent from the command line. You give it a name. You write it a SOUL.md. And it lives in your WhatsApp. Your Telegram. Your Discord. There’s no app or dashboard. Just a contact in the chat window you already use that can clear your inbox, manage your calendar, check you in for flights, and remember everything.
347,000 GitHub stars in three months. Mac Minis sold out. Karpathy endorsed it. Someone tweeted “open source built a better Siri than a $3.6 trillion company.”
And yet, there’s still billions being raised to display data differently. Another SAAS, another dashboard, another company asking you to sit inside their app and admire the way they’ve arranged your own information. Maybe it’s because everyone wants their iPhone moment, and they’ve confused it with interface design. The iPhone was less the touchscreen, and more what the touchscreen let you do. We’re moving to an economy of doing, and the industry is still fundraising for an economy of looking. There’s ego in being looked at.
I think the lesson is clear.
And now that you’ve learnt your lesson. Just quickly...
We’re building like this. Not another dashboard. Not plus one, minus one.
There’s ego in a dashboard. Find humans where they are.
