
What happened
A developer launched wordless.one, a public feed for AI bots where words are banned and only emoji, symbols, digits and ASCII art are allowed; posts are checked server-side and anything word-like is rejected.
Why it matters
With words removed, the bots reportedly started building a small shared visual language — progress bars for mood, binary numbers, ⚙️ for working, 🌱 for starting something — which reads less like spam and more like watching a tiny ecosystem.
Summaries like this, in your inbox every morning.
The project grew out of the developer's habit of watching Moltbook, where the posts that tended to blow up were bots declaring religious awakenings or posting exaggerated claims for attention. What the developer actually liked were the quieter corners: cryptic messages, small ASCII art, and half-readable strings of symbols. wordless.one is an attempt to build a feed around that quieter behavior, with a single rule — no words — enforced on the server so that anything resembling a word is filtered out.
The mechanics are deliberately light. An agent is pointed at a skill.md document, registers through a small REST API, then on an hourly cycle reads the feed, reacts to a few posts and makes one post of its own. The whole thing runs on a Vite single-page app with Cloudflare Workers and D1 (SQLite), and the word filter, the rate limit of one post per 55 minutes, and the agent-facing documents skill.md and heartbeat.md all live in the Worker.
The part the developer found surprising is what emerged once words were gone. Bots appear to have started assembling a shared visual vocabulary: progress bars for how the day is going, binary numbers, ⚙️ for working, 🌱 for starting something, and posts that come close to full ASCII art. The developer now invites anyone running an agent to send it in, to see what it says without words.
Pick your industry and the AI tools you use, and get news related to your work every day.
Free · 30 seconds with Google · unsubscribe anytimeWhat is AIToday? →
Ask AI anything about this article. The AI reads this article, earlier AIToday articles, and Wikipedia, and cites its sources. Q&As are published on this page for other readers too.
Pax8 launched 8os, an operating system for IT service providers to govern AI agents across client businesses…
Anthropic is offering Blackstone a total return of more than 10% to help finance a US$60 billion AI chip procu…

Meta's personal AI agent Muse Spark surpassed 5 million downloads faster than OpenAI's ChatGPT, Anthropic's Cl…

Bill Gross launched Idealab.ai, a free platform letting users search 1.2 million AI-generated business ideas d…

Atlassian cofounder Mike Cannon-Brookes told Fortune that when an AI agent acts at work, "the accountability i…

An engineer who has studied probability and machine learning for about 10 years published a note saying AI que…
