Weoy retweetledi
Weoy
320 posts


@uninsightful @grok explain the vc math in the tweet to me in plain English
English

the default yc round this batch (W26) seems like 4m on 40m
I remember when I first started in venture exactly three years ago (W23 batch) and most venture ppl were complaining about YC pushing their founders to do 2m on 20m
in 3 years the market went from a very begrudging 2 on 20 to a more neutral 4 on 40
interesting to think about where things land 3 years from here
English
Weoy retweetledi

@LiverpoolMTB @openclaw Peter itself has published something with the hetzet docs right
English

how I set up @openclaw on a VPS for free
no SaaS
no dashboard dependency
just infrastructure
this is exactly how I deployed OpenClaw on a Hetzner VPS
secured it properly, chose my model, and got telegram running👇
1) Spin up the VPS
Hetzner Cloud
Spec:
- 4 vCPU
- 8GB RAM
- 160GB NVMe
- Ubuntu Server
SSH in
at the time of writing, I haven't even been billed yet
2) secure it first
log in as root once
then:
- create a non-root user
- add it to sudo
- disable root SSH login
- use SSH keys instead of passwords
do not run public AI infra as root
3) basic hardening
before installing anything:
- update the system
- lock down SSH
- only expose the ports you actually need
production mindset from day one 🧠
4) install docker under your non-root user
OpenClaw runs in containers.
that gives you:
- isolation
- clean restarts
- reproducible deployment
- no pollution of the base OS
you're defining services, not installing software globally
5) deploy OpenClaw properly
clone the repo into your user directory
OpenClaw consists of:
- gateway container
- provider integrations
- environment configuration
secrets go in a .env file
inject them into the container at runtime using --env-file (or docker compose)
do not hardcode bot tokens inside config files
config is everything
6) choose your AI model
OpenClaw is a gateway, it needs a brain
you must configure a model provider
i chose claude as the backend because:
- strong reasoning
- stable outputs
- solid instruction following
without a model configured, OpenClaw is just plumbing
(model API costs are separate from VPS costs)
7) bind correctly
this catches people
inside the container, bind the gateway to lan so internal services can talk to it
when publishing ports:
- only expose what you actually need
- prefer binding to 127.0.0.1 and using a reverse proxy or SSH tunnel
- do not blindly open public ports
if it's listening on the wrong interface, it looks broken
it isn't
it's just bound incorrectly
8) configure telegram properly
you need:
- valid telegram bot token
- provider enabled
- environment variables loading correctly
- model configured
if telegram starts and immediately exits, it's nearly always configuration
read the logs
they are not decoration
9) bring the stack up
you're looking for:
- gateway starts
- model provider loads
- telegram provider starts
- telegram provider stays alive
if it survives a restart, it's properly deployed
what does this cost?
this VPS spec comes in at under €8 per month
for:
- secured infrastructure
- dockerised OpenClaw
- claude as the model backend
- telegram fully integrated
- correct network binding
- full control
cheaper than most AI subscriptions
and you actually own the stack
stay clankerish

English
Weoy retweetledi

We cracked the code on Notion AI
Here's how to get it to work 20+ minutes on complex tasks 🔥
As you know, the new Notion AI Agent is pretty insane
- multi-step reasoning
- can access databases, browse the web & get data from Slack & co
- whatever you do in Notion, the agent probably can do too
But prompting it can be a bit tricky
Initially, we'd often run into issues
For example, when handling large amounts of data, it would often start batch processing only to then pause after 10 entries hallucinating that it would continue in the background
(spoiler alert: it does not continue in the background)
But after a few weeks of trial and error, we found a few tips & tricks that greatly improve its performance
Just shared my learnings with my team on Sunday
If you'd like to see our internal doc, comment "agent" below and I'll send it your way 😇
GIF
English
Weoy retweetledi
Weoy retweetledi

Cracks me up but also makes me think about life seriously every single time! The range of emotions you'll feel reading this piece is unreal... Thank you @waitbutwhy for this amazing piece
waitbutwhy.com/2015/12/the-ta…
English

@RestaurantWill Hello sir, you mind to check again ? Just checked my dm but haven’t got it
English
Weoy retweetledi















