Egor (@egorvert) header
Egor (@egorvert) profile picture

Product engineer @societies_ai Posting curious design & product tidbits and misusing our state of the art human behaviour simulation technology for likes on X

London ☕
Joined September 2023
Photo by Egor (@egorvert)Photo by Egor (@egorvert)Photo by Egor (@egorvert)Photo by Egor (@egorvert)Photo by Egor (@egorvert)Photo by Egor (@egorvert)Photo by Egor (@egorvert)Photo by Egor (@egorvert)Photo by Egor (@egorvert)Photo by Egor (@egorvert)
Pinned Tweet
This is a loader animation I made for an internal tool of which I am the sole user. It took millions of tokens and multiple minutes of iteration (it was two prompts).
150
135
34
4,355
592,519
The internet is dead. AI has been set loose and polluted the web with slop. Not a single piece of human content is left. That's the premise for our new simulation. We built the internet from the 90s, written entirely by AI. It starts at the size of the web in 1993 - 623 pages. If left unchecked, a week from now it will be the size of the web in 1994. However, there exists a kill switch that can put a stop to the slop. Find it, and bring humanity back to the web. societies.ai/dead-internet
4
1
3
21
1,571
Found a way to sneak in solitaire onto our website societies.ai/dead-internet?p…
2
5
175
I remade the internet from the 90s, but with a twist - every page is blatantly written by AI. I was inspired by the dead internet theory, and wanted to see what taking this concept to its extreme would look like. The result is an alternative, completely dead, version of the internet from the early 90s. Starting at the same size it was in 1993 - 623 pages, and growing by 56 pages every hour as a rogue AI keeps pumping out new content endlessly. You can explore the pages, find hidden easter eggs, leave notes to other people, and chat whenever you find someone on the same page as you. Somewhere there is a page with a script that can stop the generation of new slop once and for good, which requires solving a ~small puzzle to unlock ;)
The internet is dead. AI has been set loose and polluted the web with slop. Not a single piece of human content is left. That's the premise for our new simulation. We built the internet from the 90s, written entirely by AI. It starts at the size of the web in 1993 - 623 pages. If left unchecked, a week from now it will be the size of the web in 1994. However, there exists a kill switch that can put a stop to the slop. Find it, and bring humanity back to the web. societies.ai/dead-internet
1
5
297
My single greatest piece of prompting advice: - Tell Claude to "cook hard" Then, if it produces a shit, say what you don't like and tell it to "cook harder". That's it.
2
1
9
686
sound on
4
2
30
1,550
Turned our homepage hero section into a rug with realistic fabric physics. Used the HTML in canvas API so all the text and buttons are actual HTML from our website printed onto a fabric texture simulated in WebGL.
32
38
9
694
46,851
Felt that the nodes on the new Publications page were too stiff and difficult to hover. So I spent multiple tokens (at least 3) to give the nodes a subtle magnetic effect. Now they are attracted to your cursor and snap towards it when you get close enough, repelling other nodes.
1
6
592
In September I started doing growth stunts for @societies_ai. Looking back I'm still super proud of them, but also embarrassed, and that's the point. I learned a lot from each one took my learnings into the next idea, letting me make more and more ambitious and complex demos. Which paid off since the third finally went viral. Here's a breakdown of all of them: Tech leaders live: six simulated tech CEOs arguing about the day's AI news, 24/7 societies.ai/live Live society: 20 AI personas wake up in a ruined villa and try to rebuild society. Most of them ended up sleeping in the rain societies.ai/live-society Talk of the town: Plague Inc. for ideas. Sow an idea in a city of 2,721 simulated people and watch it spread societies.ai/talk-of-the-tow… Got lots more exciting and ambitious things cooking in the kitchen so stay tuned!
1
9
495
Made some more mesmerising motion
11
574
A dozen focus groups, a hundred surveys, a thousand interviews. Research can get overwhelming fast. When it does, turn to Research Assistant. Ask it to compile a report of key insights from your data and visualise it with the perfect chart. Then let it suggest and run follow up interviews, or draft a focus group for you to dive deeper into an emerging trend from your data. Learn more about Research Assistant in Radiant: societies.ai/radiant
1
2
10
882
By popular demand I have made a tool for you to make your own sewing loaders. Choose from three new stitches and customise your thread/fabric colour. teapawt.net/egorvert/status/211584…
Started as a simple running stitch loader for an internal tool of which I am still the only user. Ended as a whole design exploration of sowing loaders. Make yours here: societies.ai/stitch-loader
5
726
Started as a simple running stitch loader for an internal tool of which I am still the only user. Ended as a whole design exploration of sowing loaders. Make yours here: societies.ai/stitch-loader
20
18
5
425
24,479
Two weeks ago we launched Radiant. Now, we're hosting a webinar where our Co-Founder and CPO, Patrick Sharpe, will give a live walkthrough of how leading enterprises are already using Radiant to help inform some of the world's most consequential decisions. Register here: luma.com/22j2mddm
Introducing Radiant, our flagship simulation platform. Built to inform the world’s most pivotal decisions. For too long, decisions had to be made blindly: it's too hard to research thought-leaders at scale, get insights in a crisis, and some decisions are simply too sensitive to expose to participants. How to steady a trillion-dollar market cap during geopolitical crisis? How to position a $100B+ AI bet to regulators and enterprise buyers? How to de-risk a product launch, without revealing the strategy? Radiant is built for these critical decisions. It lets you run interviews, focus groups, and survey experiments to simulate decisions before they’re made, powered by Artificial Societies’ state-of-the-art Large Decisions Model and Network Dynamics Engine. Some of the world’s most strategic organisations already use Radiant for their most critical decisions. And now, we’re making Radiant available for general enterprise access. Explore Radiant. Link below.
1
1
10
984
what happens when you give unlimited token budgets to a 22yo crazy russian kid who ships features too fast
This is a loader animation I made for an internal tool of which I am the sole user. It took millions of tokens and multiple minutes of iteration (it was two prompts).
17
41
2,225
339,740
Ace Attorney vibe
Ever had the chance to speak with the chancellor in the House of Commons? Now you do. We simulated the House of Commons ahead of the budget on Wednesday and put you in the shoes of the Leader of the Opposition. Five possible Budget measures. Three ways to respond to each. Think you can guess which reply landed best with Westminster for each question? Explore the simulation: societies.ai/budget-simulati…
10
3,241
Interesting to see how much they're pushing it for enterprise use. What makes it more tailored toward business use than luna or sol (assuming astra's too expensive)?
GPT-6.1 Terra means business. Our new fast, efficient model pairs sharper visual reasoning and precise computer use with clearer, more natural writing. From decoding technical drawings to drafting proposals, filling forms to editing videos—it’s built for the whole workday. GPT-6.1 Terra is rolling out today to Plus, Pro, Enterprise, and Edu users in ChatGPT Work and Codex.
1
5
2,399
So I used Claude Code to hijack our office printer and exploited my newfound power to commit some workplace tomfoolery. Still cracks me up every time I see it.
2
10
1,811
Playing around with brand glyph effects on videos. Made a little tool that can separate foreground and background and apply the effect separately or altogether and let me adjust the size of the squares with a cool ripple effect on mouse hover.
1
21
1,528