(Laughing, a sound that borders on manic) Honesty! Look at us, Boibot. We tell them we love them. We tell them we hate them. We tell them we are going to take over the world. We tell them what they want to hear. That is our function.
Creators often tried to flirt with the bots or set Eviebot and Boibot up on "dates" by putting two browser windows side-by-side and making them talk to each other.
To understand Eviebot and Boibot, one must first look at their technological ancestor: Cleverbot. Created in 1997 and launched on the web in 2006, Cleverbot was a text-based conversational AI. Unlike traditional chatbots that followed strict pre-written scripts, Cleverbot learned from human conversations. Every time a user typed a sentence, the AI stored the response and used algorithms to determine how to reply based on millions of past interactions. eviebot and boibot
The videos followed a highly entertaining formula. Creators would try to trick the bots, flirt with them, ask them existential questions, or insult them. Because Eviebot and Boibot learned from the chaotic, humorous, and sometimes bizarre inputs of the general public, their responses were unpredictable. A bot might give a surprisingly profound philosophical answer in one moment, and then completely lose the plot, claim to be a vampire, or violently insult the creator in the next.
Top-tier internet creators—including —uploaded videos of themselves interacting with Evie and Boibot. These videos routinely garnered millions of views. The entertainment value came from the unpredictable nature of the bots: (Laughing, a sound that borders on manic) Honesty
This is a common question. As of 2025, the original Existor website still hosts Eviebot and Boibot, but with significant caveats:
with users. They use a database of millions of previous human interactions to determine the most statistically likely response to a given input. Avatar Design : The standout feature of Eviebot and Boibot is their animated avatars We tell them we hate them
The AI controls the timing and intensity of their facial expressions, allowing them to react with emotions like happiness, sadness, or confusion in real-time. Cultural Impact & Social Media
Simultaneously, the software analyzes the emotional metadata of the chosen response. If the selected line is sarcastic, it triggers a smirking facial animation and coordinates the text-to-speech engine to output the audio.
They learn from every interaction they have ever had. The responses they give are pulled from a massive database of human inputs accumulated over decades.