According to WIRED, the hottest new AI chatbot in San Francisco is not powered by a large language model. ChatTJB, the creation of 32-year-old Tucker Bryant — a public speaker, artist, and former project manager at Google — works because Bryant sits down and answers each prompt himself. WIRED reported that as of August 6, ChatTJB had received over 30,000 queries, with a peak of about 5,000 prompts per hour.
A $6,000 Billboard Sparks Viral Growth
WIRED reported that Bryant launched ChatTJB in April, but it did not take off until July 27, when his $6,000 billboard went up in San Francisco. The billboard describes ChatTJB as "the leading chat interface powered by AI," where AI stands for "average individual," according to WIRED. News coverage followed soon after.
How ChatTJB Works
Bryant did not train a language model to answer in his style. WIRED reported that he literally answers prompts by hand. For image requests — such as a skunk wearing sunglasses — he scribbles a drawing. For a request like how to fry an egg, he offers a suggestion. Maybe. WIRED noted that when it asked how to become a morning person, no response had arrived at the time of writing; the website states that "thousands of people are writing in right now."
Bryant considers ChatTJB an art project "first and foremost," according to WIRED. He is not anti-AI — he told WIRED he vibe-coded the ChatTJB website using the tool Lovable.
The 'Cognitive Surrender' Concern
Bryant's underlying worry, WIRED reported, is "cognitive surrender" — accepting AI outputs with increasingly little critical thinking or scrutiny. He described asking a large language model whether to wear short or long sleeves on a 65-degree day in San Francisco, despite living in the city for about seven years. "Like, I know how to handle the weather. And yet, it felt like an easier thing for me to defer to a LLM for advice on that," he told WIRED.
WIRED reported that many users first treated ChatTJB as a joke, but then shifted to personal, sincere, and genuine questions when they realised a human was responding. Over the past couple of days, Bryant has drawn a frog eating spaghetti, San Francisco mayor David Lurie as Godzilla, and two black cats eating grass on a table.
Traffic, Volunteers, and Monetisation
WIRED reported the following metrics:
| Metric | Value |
|---|---|
| Launch month | April |
| Billboard cost | $6,000 |
| Billboard went up | July 27 |
| Queries by August 6 | Over 30,000 |
| Peak prompt rate | About 5,000 per hour |
| Volunteers applied | Over 3,000 |
| Volunteers selected | 10 |
| ChatTJB Pro price | $5 per month |
| ChatTJB Pro subscribers | About 4 |
Bryant told WIRED he expected the project to last only a couple of days, but plans to continue after seeing people's enthusiasm. He has enlisted 10 volunteers from a pool of over 3,000 applicants. ChatTJB Pro, which WIRED reported the website calls "a worthless tier," costs $5 per month and offers nothing but the knowledge that users are supporting the art; about four people have paid so far.
Bryant told WIRED:
The way I'm interpreting it is that the value for them is just getting a human to bounce those questions off of, as opposed to getting super expert advice.
He described ChatTJB on YouTube as "the worst LLM of all time," according to WIRED.
What the Experiment Shows
The ChatTJB case, as reported by WIRED, puts a human back into the AI loop at notable scale: 30,000 queries and a peak of 5,000 prompts per hour. For enterprise technology decision-makers, the explicit concept to assess is "cognitive surrender," which WIRED defines as accepting AI outputs with decreasing critical thinking. Bryant's own weather-advice example — deferring to an LLM despite years of local knowledge — is the same dynamic that can affect employees relying on AI-generated recommendations. WIRED's report shows that users responded differently once they knew a person was behind the interface — a documented behaviour in this experiment.
Sources:
