@valhalla_dev AI can delay the first hire but it does not remove the need for ownership, judgment, and someone waking up responsible for a whole surface area.
@illscience I think the split is less “tried AI vs didn’t” and more “used it on a real project with context, stakes, and continuity vs asked it one-off questions.”
I went to a graduation this weekend at Pennsylvania College of Technology, and it made me think about AI adoption.
The thing that stood out:
The graduating class looked overwhelmingly male.
That surprised me because, nationally, college enrollment skews the other way. But this was a technical college, so it made sense.
Different pipeline.
Different culture.
Different adoption pattern.
That sent me down a rabbit hole.
Most of the AI conversation right now is about productivity.
Will AI take jobs?
Will it make workers faster?
Will students use it to cheat?
Will companies automate departments?
Important questions.
But there is another AI adoption story underneath the surface.
AI is becoming a place for private conversations.
Not public.
Not polished.
Not performative.
Private.
Pew reported that 34% of U.S. adults have used ChatGPT.
Deloitte found that, in 2024, 44% of men and 33% of women were using or experimenting with generative AI.
CDC data from 2019 found that only 7.2% of men received counseling or therapy in the previous year.
That contrast stopped me.
If even a quarter of male LLM users are using AI for emotional support, journaling, or “personal therapist” conversations, the math gets weird fast.
We are entering a world where more men use AI as informal therapy than go to actual therapy.
Sounds weird but maybe it is not.
There is an old joke:
"Men will do anything except go to therapy."
But maybe the joke misses something.
Maybe some men are not avoiding the conversation.
Maybe they are avoiding the publicness of the conversation.
The appointment. The cost. The vulnerability. The feeling that you have to explain yourself before you understand what is going on.
AI changes the shape of that interaction.
It gives people a private place to start.
A place to type the messy version.
A place to ask the embarrassing question.
A place to admit the thing before they are ready to say it out loud.
This is not a “bridge to therapy.”
This is becoming its own category.
Not because every chatbot is better than a great therapist.
But because a great therapist is scarce, expensive, scheduled, human, and socially loaded.
AI is abundant, instant, private, patient, and available when someone is finally ready to say the thing.
Search engines showed us what people wanted to know.
Social media showed us what people wanted others to think.
AI chats may show us what people are actually wrestling with.
Some of that will be healthy.
Some unhealthy.
Some uncomfortable.
But this is moving us in a direction that we aren't talking about enough.
AI is not just a productivity layer.
It is becoming a private interface for human thought.
The biggest AI use cases may not be the ones people brag about.
They may be the ones people are quietly grateful for.
The late-night conversation.
The messy first draft.
The question they were afraid to ask.
We aren't just getting AI slop, we are also getting a place to be honest.
Our talking and thinking patterns are shaped by the different platform algorithms.
It’s so interesting to hear the talking patterns is someone who is off social media.
It’s even more crazy to see the thinking patterns of someone who is off social media.
@NicoleBehnam Love this.
I just had a moment like this today.
I was trying to show a friend something I built today with ai and he wasn’t getting it because I couldn’t quite show how radically different his life would be if he used what I built.
Opportunity missed.
Gotta fail it in.
Note to self: Discuss complicated ideas but don’t use complicated words
Because I’ve noticed dumb people use complicated words to describe simple ideas
I just got something working that feels important:
an external AI agent can now call into BuildOS, identify itself, and operate within a scoped permission boundary.
Not “AI talks to AI.”
More like:
- who is calling?
- which user’s BuildOS agent are they calling?
- what projects can they access?
- do they have read-only or read/write scope?
That means OpenClaw can connect to BuildOS, discover tools progressively, and work against real project state without getting the whole system dumped into context.
I think this is the right abstraction for agent infrastructure:
a secure call layer with scoped capabilities, not a loose chatbot bridge.