After school, Karandeep Anand often finds her 6 -year -old daughter in a conversation with a Ai chat while eating snacks on the kitchen counter. He is too young to write, he only has his own account to characters.
Character.AI is a startup from AI colleagues (though Anand likes to say he is a Play-Play Ai Startup, which we will enter later). He took over the post of CEO in June in the midst of a potentially devastating demand for his parent company and the questions about child safety. When I ask him if he is worried about his daughter who connects with a Chatbot Ai instead of a true human, he is quick to say no.
“It is very rarely, in any of these scenarios, a true replacement for any human,” Anand told me during a video call later last week. “It is very clear to the application that, well, this is a play of role and entertainment, so you will never start to deepen this conversation, assuming that it is your real companion.”
It is a delicate time for the character.
Last August, Google entered with an approximate agreement of $ 2.7 billion to graduate character technology. As part of the agreement, the two character co -founders. AI went to Google’s AI Division.
Anand, who previously worked as Vice President of Business Products at the finish line, was commissioned to collect the pieces, which he did in part by leaving behind the founding mission to deliver the personalized superintelligence to focus on Ai’s entertainment.
“What we resigned was this aspiration that the founders had of the creation of AGI models; we no longer do it. That is, the hundreds of billions of dollars of the investment struggle, which Big Tech struggles,” says Anand. “What we obtained in return was clarity and focus, to be able to uniquely follow the vision of the entertainment ai.”
As part of this change in strategy, character, and no longer seeks to build his own border models. “In the last six months, we have done a lot of work to get off our text owners and start using open source models,” says Anand. The company has tried a few: Met’s Llama, Alibaba’s Qwen and Deepseek. “Open source models are exceeding the hands of the property model,” says Anand.
Running an AI startup without billions of dollars in revenue can be a brutal equation, and the character is still finding out how math works. The company told me that it is generating revenue at a rate of more than $ 30 million and goes to the road to achieve $ 50 million revenue at the end of the year. When I asked Anand how many users pay $ 10 monthly subscription, he did not give any number, but I said that “monetization was not a focus until four or five months ago.”
“Since I have been on board, it is very clear that we need to monetize us.
