[ad_1]
OpenAI introduced @ mention capabilities for custom GPTs within GPT-4, changing how people interact with the generative AI chatbot and making the platform more enticing for publishers.
Previously, people could ask ChatGPT questions like, “Stylish summer attire for men,” and receive a generic array of responses based on GPT’s existing training data. Now, the @ feature lets people specify their preferred custom GPT, like @Vogue, to the query and receive a tailored response. For publishers, this lets them index their custom-made chatbot from the GPT store within the latest ChatGPT.
“It’s not a new user acquisition tool, but a way to make it more convenient for existing users of your GPT to engage again,” said Josh Jaffe, Ingenio’s president of global media. “OpenAI could easily update this tomorrow and make it into a discovery tool.”
Publishers face diminishing search and social traffic. Some are turning to chatbot-driven chat channels, such as Quora’s Poe, for discoverability and referral traffic. Other publishers are actively exploring the launch of chat channels on Instagram direct messages, which specifically target followers who comment on their posts, sources told ADWEEK.
The integration feature was first spotted last Friday by Dan Shipper, CEO and co-founder of AI business newsletter Every, and it is currently available to limited users.
Facets of the @ feature are still unclear—will custom GPTs appear alphabetically, will it only show GPTs people have interacted with before, or will it be algorithm-based on user data? Jaffe noted that so far, only the custom GPTs he had interacted with previously showed up.
“[The @ feature] could increase the amount of interest for a publisher to be available in [GPT’s] ecosystem,” said Nick Martin, co-founder and chief commercial officer at Direqt. “There’s more of an opportunity to get in front of people.”
OpenAI’s GPT Store launched earlier in January and costs $20 monthly: 100 million people used ChatGPT weekly, OpenAI CEO Sam Altman said during its developer conference in November.
But unclear monetization models and the potential for chatbots to replace people visiting publisher websites (where they are monetized more effectively) leave publishers weary of investing in custom GPTs.
