Passing by the Solid coffee shop downstairs from the office, I was shocked to find that the original storefront was empty and deserted.
I thought it was another somewhat sad story.
But as I followed the notice on the glass and arrived at the new storefront just 30 meters away, my mood suddenly brightened in this rainy weather: the familiar owner was still there, the new shop was larger, and about ten more seats had been added.
People still need in-person communication.
This might just be an incidental example, but it's a question I've been thinking about for over a year. Suddenly finding evidence in familiar life provided double the emotional comfort.
Even after a weekend of cooling down, I believe everyone's attention on Manus will remain high this week. That dream of achieving full automation through model capabilities is rooted in everyone's heart.
Whether through "Computer Use" or LAM (Large Action Model), completing "work" automatically like a game bot or script is what many expect from AI—and it's also the "threat" many feel from it.
On one end, there is intelligence and automation that seems increasingly "unstoppable."
But on the other end, our emotional demand for face-to-face interaction seems to be growing stronger. After years of being accustomed to ordering via apps and quickly picking up coffee without staying, a coffee shop that adds a dozen seats (the original shop had a few seats, but they were mostly for "very frequent regulars" chatting or customers waiting for coffee) seems to provide some different answers.
This is probably a "fold" I've been thinking about for a long time: one side is getting smarter and faster, while the other is getting more human and slower.
In the second half of the mobile internet era, people found a model called O2O, Online to Offline.
Perhaps in the second half of AI, we will also find a model. For now, I'll name it I2H: Intelligence to Human.