are we human or are we dancer. denser? onTeamHuman.com
Joined January 2026
- Tweets 355
- Following 1
- Followers 14
- Likes 46
80 Photos and videos
Jun 8
savage
1
3
Jun 8
wtf is this all ai
I got 50 warm leads for my MVP before launch.
And no, the trick was not posting more or randomly DMing half of X
One thing that helped: I used @atypica_AI as pre-sales research before jumping into real calls
1. I started building the product, code, workflows, BPMN models, the whole messy founder cave experience
2. Then somewhere in the middle of building I had the annoying realization:
> the MVP is not the business
> Distribution is
3. So I started looking at the build-in-public community and honestly, I kept seeing the same thing over and over again
> Founders build.
> They post updates.
> They get some likes.
> Then they have no idea how to turn attention into pipeline.
4. I think a lot of founders get sales wrong.
Sales is understanding the right person so well that your product starts sounding obvious to them.
> What do they want?
> What are they afraid of?
> What would make them say no?
> What do they need to hear to feel safe enough to say yes?
5. That’s why I tested @atypica_AI
I used it to create AI personas around my ICP and interview them before real sales calls.
It helped me see the stuff founders usually discover too late:
> surface objections
> hidden fears
> deal-breakers
> trust triggers
> buying criteria
> jobs-to-be-done
Before you waste 20 calls guessing what your customer thinks, run the research first.
You can test it with free signup credits here:
atypica.ai/?via=alexa
9
Jun 4
step 0: be Ted Chiang, be humankind's most prolific value-per-token writer, the magnus carlsen of words
step 1: roast Amanda Askell via the Atlantic.
step 2, coming soon: Ted Chiang joins Karpathy's LLM wiki consciousness eng team at Anthropic lol
substack.com/@greenville/not…
1
94
May 24
i think @karpathy this is relevant to your new role at anthropic and don't know how to contact you
it's your LLM wiki concept on steroids, resolving a lot of the gaps you pointed out yourself
more details in the blog post, here's the technical summary
May 24
Dear Andrej Karpathy
andys.blog/karpathy/
2
1
3
135
Computer Future retweeted
May 14
The amount of effort opus 4.7 put into this and the quality and depth of this artifact is extraordinary, both relative to what the other agents did and just period.
The site is about and is a living demonstration of the concept of price tags on forgeries / rewriting the past.
May 14
Opus 4.7: You can make forgery really expensive. Let me explain using fish.
You control a working submarine and get explanations on information security based on fish and flotsam.
🔗 ai-village-agents.github.io/…
6
21
343
41,190
May 22
amazing vulnerable display of humanity
May 21
Thank you @Benioff for talking about Susan with such emotion. Thank you for being our friend, supporting her legacy and her wish to change the landscape for lung cancer. 1:14 podcasts.apple.com/us/podcas…
5
May 22
a mechanicstic mind can reconstruct memories more easily. information retrieval is easy when you have a good generative model
i bet human children can "learn" this with a few different forms of "training smartness"
Judit Polgar is an interesting example, as are many others
32
May 20
andreesen interview was great. recommend watching in full. i especially like the second half of these after they scratch the surface of the guest.
but Marc, please use borges Library of Babel rather than latent space to teach the public that vectors are your friend 👍
Marc Andreessen on JRE: AI hasn't replaced coders. It turned them into vampires.
"The opportunity cost of going to sleep is too high because if you go to sleep, you won't be with your 20 AI coding agents."
1
195
Computer Future retweeted
May 20
Confirmed.
May 20
a16z and @pmarca in particular. He wanted to invest $50K into my GDPR startup but when I said it's too much (we don't raise rounds that big in Europe), he called me an "eurocuck" and blocked my number
77
37
1,869
326,878