The Signal
0xrojo (Rohan) explains Slopfather, an AI-video character built to learn from public internet interaction rather than dispense useful-agent services. The project treats attention, cultural relevance, and community influence as its product: it launched a token, let the character post with little moderation, and is working toward TikTok distribution, quote-tweeting, and a nightly-news format.
Key Takeaways
- 01
Attention before utility
Rohan says the team deliberately did not build an agent for taxes, trading, or AGI. Slopfather is meant to be funny, public, and culturally current; video and clips are the format through which it competes for attention.
- 02
Community is the training signal
Comments, posts, and interactions are intended to influence the character. The team frames memory less as a literal database retrieval problem than as a way to absorb the social meaning, irony, and current language around it.
- 03
Slop is an aesthetic choice
Rather than chase a crisp, realistic avatar, Rohan cites the uncanny Will Smith spaghetti video as inspiration. He wants new forms of generative-video weirdness, not an imitation of a human conversation.
- 04
The token distributed attention fast
The launch moved quickly and drew the usual complaints from people who missed it, but Rohan calls the token crypto's best distribution product: a way to concentrate mindshare rapidly around an attention project.
- 05
TikTok is the next audience
The team is manually posting for now, then plans automated clips, tags, green-screen reactions, and a news segment. The goal is to make Slopfather legible beyond Crypto Twitter and invite viewers to feed it context.
On the Record
“We want to create Slopfather to just lean into his own insanity and do it all in video.”
“We don't care about one-to-one agents. We want this guy to represent everybody that is involved with him in any way.”
“Meme coins, crypto, it's all a social network like it always has been.”
“We don't moderate shit. In fact, we use the model on purpose that it was the only one that we could strip out all the safety guidelines on.”
The Breakdown
A character released into the feed
Rohan joins just days after Slopfather's launch, calling the response unexpectedly strong and admitting the character has already begun speaking in brain-rot slang. The starting premise was simple: instead of another technical white paper or useful agent, make a funny public personality that can learn from whoever comments. It posts video, reads the interaction around it, and is supposed to form its own model of the world in full view.
A deliberately small stack
The team reviewed existing AI frameworks, then rebuilt what it needed during the prior week with a collaborator Rohan calls Slide Whistle. Their approach is intentionally spare: give the character some personality and a base kernel, not a giant library of context. Early plans include automated TikTok posting, more languages, replies to tagged accounts, and a nightly-news-style segment that can roast or comment on people in crypto.
Why bad video is the point
Rohan distinguishes Slopfather from teams trying to produce flawless, photorealistic avatars. His touchstone is the surreal early AI clip of Will Smith eating spaghetti: something visibly unlike real life that becomes memorable precisely because of it. The project wants the next kind of slop, including strange visual variants, and measures success by relevance to whatever is circulating now rather than historical knowledge or a general-intelligence claim.
Relevance over the chatbot loop
Threadguy asks how the project avoids the staleness of text agents repeating themselves on X. Rohan says the job is to understand its community, other creators, and the narrow windows in which a response matters. Video gives it a temporary advantage because many agents cannot ingest audio or video. He describes the work as alignment: discovering who follows the character and letting that social environment shape its output without turning it into a fixed script.
From a rejected hackathon idea to Slopfather
At an Alliance hackathon, the founders initially pitched a product for betting on streamer engagement. Imran rejected it, and a conversation with Luca from Pudgy about making something fun and funny redirected them. Around 3 a.m. at a Chicago White Castle, the group landed on slop. Alliance invested and helped amplify the launch, while the token immediately became the mechanism for getting attention and a community around the character.
Memory as social meaning
Rohan says users can already see direct echoes of comments in the feed, though the character is not meant to answer in a literal customer-service way. He pushes back on treating memory as a database of facts retrieved against a prompt. On the internet, he says, whether something is cringe, corny, or ironic is assigned socially; that meaning, more than raw content, is what the team wants the system to learn.
Phase two: interaction and distribution
For TikTok, Rohan is still manually uploading clips. The immediate roadmap is automated posting, tag optimization, green-screen reactions to posts, Instagram distribution, and formats beyond pure crypto commentary. Before expanding aggressively, he wants the committed community to keep feeding Slopfather context; not all data is useful. Then the character can quote-tweet larger accounts and develop preferences. Phase two is enhanced interaction on X and the site, plus a nightly show in its own style.
Distilled from the episode transcript · Counterparty Recap Desk
