Emails now arrive addressed to 'Tomasz or Tomasz's agent.' Venture capitalist and writer Tomasz Tunguz documents the quiet normalization of human-AI communication proxies, where senders no longer assume a person will read, think, or reply. This is not a hypothetical future. It is current inbox behavior.
The piece catalogues the specific mechanics of this shift: Gmail's autocomplete replies ('Sounds good!'), AI customer support voices indistinguishable from human ones, and ElevenLabs cloning a voice from 30 seconds of audio, including the ums and pauses. A friend sends voice memos to prove he is real. The verification problem is already here, and the workarounds are already failing.
Tunguz's sharpest observation is buried near the end: adapting to AI intermediaries is not rudeness, it is rationalism. The readers who will get the most from this are not the ones who want a conclusion about AI identity. They are the ones who want to understand why the polite social default is shifting toward assuming the robot first, and what that assumption costs.
[READ ORIGINAL →]