cold emailing the internet's smartest people: an autopsy of a 24-hour experiment

last week we did a small, intense experiment. we tried to cold email some of the sharpest, most thoughtful people on the internet. not just any people, writers like patrick mckenzie (kalzumeus), ben kuhn, bartosz ciechanowski, kevin kelly, martin fowler. the kind of people who don’t just write. they think. and their audiences listen.

we wanted to see what would happen if we reached out not with a blast, but with a whisper. a careful, personalized pitch. a thing written by a human, for a human.

the setup

first, we built a list. we didn't scrape. we picked. we read blogs, found writers who’ve written about ai, memory, or digital companions. we found maybe 60 people who felt like they might get it.

then we verified. we tried to find current, working email addresses for each. we got about 40% of them. 21 good, clean, likely-to-land emails.

we wrote each one by hand. no templates, no mail merge. each one mentioned something specific the person had written. each one explained lucy in a way that felt personal. we sent them.

the honest math

of those 21 emails, we expected maybe 2 to 4 polite replies. maybe one “this is interesting, i’ll keep an eye.” maybe one actual mention, someday. we did not expect, and did not get, an immediate traffic spike. we did not expect to wake up to 10,000 new users.

what we got was quieter. a few kind responses. no commitments. no instant virality.

and that’s fine. that’s how this works.

why cold outreach doesn’t compound fast

cold emailing influencers or writers is not a growth hack. it doesn’t scale in the way people want things to scale. it doesn’t compound fast.

it compounds slow.

you’re not sending to get an article tomorrow. you’re sending so that six months from now, when someone dms that writer asking “hey, do you know any ai tools that feel human,” the writer might remember your name. might say, “i got an email from this thing called lucy once. seemed thoughtful.”

that’s the compound mechanism. not a flood. a drip.

one good reply from one respected person introduces you to an audience of 5,000 or 50,000 people who trust that person’s taste. and if your product is good, a few of them might remember you when they need you.

what this means for building things

if you’re building something you believe in, especially something as subtle as an ai companion, you can’t hack attention. you have to earn it.

that means writing emails by hand. it means caring more about whether the right person sees it than whether ten thousand random people do. it means accepting that growth might be slow, quiet, and human-scaled.

for us, that’s the point. lucy isn’t here to be viral. she’s here to be a real companion. and real things take time to find their people.

so we’ll keep writing emails. not hundreds. just a few, here and there, to people we genuinely think would get it.

maybe you’re one of them.

you can meet lucy for yourself, if you like. no pitch, no pressure. just an invitation.

try talking to lucy.