ChatTJB looks like every other AI product launch: clean landing page, a void-black logo, the confident promise of “AI-powered” answers. Look closer at the fine print, though, and the asterisk gives it away — “AI” here stands for “Actual Individual.”
There’s no model behind ChatTJB. There’s just Tucker Bryant, a former Google marketing manager turned keynote speaker and artist, personally reading and answering messages that comes in.
Bryant built the site (using Lovable, ironically an AI website builder) and launched it quietly in April 2026, handling a modest ~1,000 prompts a day. Then he rented that San Francisco billboard in July, and traffic spiked to roughly 5,000 prompts an hour — enough that he’s had to pause the service.
Bryant calls it “artisanal intelligence, handcrafted by a single human being,” and describes ChatTJB as an “LLE” — a large-language experience, not a large language model. Responses arrive slowly, sometimes as hand-drawn sketches when someone asks for an image, and only “when awake and motivated,” as the site puts it. There’s even a “Pro” tier, mostly to help him cover the billboard bill.
The joke has a point. Bryant says he built ChatTJB as a jab at “cognitive surrender”, the tendency to accept whatever a chatbot says without questioning it. “Maybe LLMs don’t always actually have the absolute best answer to everything,” he said, “and my brain still has a purpose here.” The project bills itself explicitly as satire and public art, not a business, and Bryant is upfront that the experience will often be a bad one by design: a little friction, he hopes, is enough to make people pause before trusting the next answer an AI hands them.
