ChatTJB: a service where a real person replies instead of AI went viral after a San Francisco billboard

Stanford graduate and former Google employee Tucker Bryant has launched an unusual service called ChatTJB that mimics interacting with a chatbot, but all replies to users come from a real person. The project is intended as satirical art rather than a competitor to artificial intelligence systems.

Bryant, who now works as an artist and speaker, put up a billboard in San Francisco costing $6,000 a month and answers roughly 1,000 queries daily. The service's description says ChatTJB is a "next-generation single-operator large language experience project" where instead of sampling from a probabilistic model, each request is processed through a "biological thinking substrate." In practice, this means a user asks a question, and Bryant reads it, thinks it over, and replies when he wakes up and feels inspired.

The idea came after his partner introduced him to the concept of "cognitive capitulation" — a state when people blindly trust chatbot advice. Bryant realized he had experienced this himself and created ChatTJB as a way to bring awkwardness and unpredictability back into chatbot-like interactions.

At first, requests were scarce due to the billboard's poor location, but once its image spread online, the author found himself flooded with questions. Many visitors turned out to be trolls, which Bryant took in stride, but there were also surprisingly touching messages. One user wrote a couple of days later: "It's my first night of my honeymoon and I don't feel relaxed, is that normal?" — at that point the author switched from a super-random demeanor to a supportive friend.

Users often reminded Bryant to take breaks and drink water, which was helpful since on some days he did not take good care of himself. This prompted the author to consider expanding the project — he plans to let some approved volunteers help with replies.