Eating pizza and building software at @internetstudio.bsky.social

I am glad vmax.ai unveiled the world document designs today, as our research expands and our learnings develop, I invite you all to come along for this world building journey I am leading! Every day I feel lucky to be in the presence of so much cool research 💯

We love hosting DWEB Seattle

Soon we'll start unveiling all the cool work we have been doing for the vmax.ai world document model I have been crafting. I hope you learn from it and take from it and share the awesome research you discover from us 🧱

mood

New update that brings nicer connections with bezier curves and waypoints, plus a few bug fixes

Added in open source fonts from google and working on making it easy to work with text on the page

I ended up having to throw out two days of attempted work, I still think the majority of original interface work needs to be done by hand. There are just too many hallucinations in 0-1 design work, I’ll probably only use agents when I am sleeping and when I am done with patterns

still going, still dreaming

Should designers use /loop?

If you know you know.

I just hope Claude or OpenAI used our open source front-end components and CSS in the training data or whatever so I can feel useful in grand transition we are swept up in

I just want to build world class tools, build great companies and fund great companies, and thats what I think is a good time. I have felt this way for the last 20 years and I am finding the optimism that the best is yet to come, and all I need to do is to keep doing.

my /loop experience in CC - I made a 3k line plan - I have a harness “NEXT” doing 100k token each run - I said /loop 1m take full control of this workspace, execute NEXT - NEXT has a stopping point measure I started it, slept, it had ran for 8hrs, pretty much 100% completion

Without hyperscaler compute subsidies, Claude Max 20x alone would probably cost closer to $2,000/month than $200/month

All my software engineering pals just need survive 2 more long years, relearn our whole development toolchain, be great at graphic design and systems design and we should be able to get through all of this hype and wage shock.

a world for every document

Version 2.9.0 of the VMax World Document introduces logging, and sets the stage for agents to start building things in their world 🧱

Co-working day with the ladies 💻📚🌿🫶

AI is definitely not taking everyones jobs. Its just taking away the tasks from skilled people who made careers from doing those tasks. That is a small population of people, its collateral damage and we also saw this coming. No one was going to have me write perfect CSS forever.

you can now spend your dots on upgrades. full changelog in game.

I put extra attention into the world document for making sure its agents could pursue tasks, way-find (A*) well, and find their way through worlds. I think all of that will make incredible scenes and many possibilities, in your browser

I love pairing worlds with writing, looking forward to the future after pushing like hell on the latest release of my world document.

I am grateful every day that I get to design and ship platform and websites. My best work is yet to come and if the music stops, I am still pushing.

If you mention humanity too many times in your white paper I will render an active volcano

For the VMax World Document System, I have created maybe one of the first agentic scenes with a dormant volcano… guess what comes next?

We love throwing pizza parties for the local internet builders cause thats a little much needed happy compared to whatever else we talk about these days Thanks for sponsoring us 🤝🫂

Lush forest island fog

Debugging empty posts, a simple desert island

Using a white paper to also generate a relevant world is my dream for the VMax Document System

For the VMax Document System, very close to perfecting the scene generation for capture the flag, and I have so many other ideas now that the physics and agent behavior are refined ⚔️

Sometimes you got to find your party, so you can get ready for whats next

Maybe a week away from demonstrating the world document editor capabilities and its ability to generate real time strategy scenes based on post content.

For my VMax world document editor, it was clear to me that for world class research, the scenes that formed for the topic needed to pair well Here you can see a simulation of agents leaving their boats and encountering a Castle based on doc content I can’t wait to show you more

The best investment you can make is in remarkable people

dots now start to merge together when players accumulate large amounts of dots, making it possible to reach much higher scores

Touch grass and punch marbles

I am building a world and is the infinite jump glitch a feature or a bug?

I am building a world and this level just takes a little time and practice

vastly improved the style and handling of the connection ports

If you write nothing into the document, your island is barren and covered with sand. Desert power

plumbing bytes between terminal nodes

I built Sacred Computer by hand (the old ways lol) so that it would be easy to build terminal UIs that match in the web context and CLI context. Today version 2.0.0 has agent skills that make it easy to port Sacred Computer anywhere. Just tell your agent to read the page

Our friends at @internetstudio.bsky.social are running a small program to help builders ascend into being founders/partners. Here is a video of studio partner @elij.dev, who often helps take products from zero to one for many clients. He made diagram.page on his own without AI!

Its a world, a long form document, a biome, and sometimes a few raindrops

In my document builder that also does world generation, I added lush forests, and also kept the way to burn the whole document down. Kind of like burning the world down 🫠

Based on what you write, sometimes the world turns into a green paradise

You got to keep on moving!

I am building a world, and there is a castle and flags

I am building a world