(This column originally appeared in Forbes)
While the mainstream public has been obsessed with ChatGPT, the conversational AI chatbot that’s been popular for making recipes, generating blogs and writing jokes, there’s another AI robot that’s creating even more excitement in the AI world. It’s called AutoGPT and it will have an enormous impact on your business in the next few years.
AutoGPT is based on the underlying technology behind ChatGPT. But its biggest difference is this: ChatGPT relies on humans to provide it guidance. AutoGPT doesn’t. What does this mean? Let’s take an example using your Customer Relationship Management system.
Let’s say you want to create a new marketing campaign that chooses everyone in your database with blue eyes, green hair, who is also a Philadelphia Eagles fan and a buyer of one of your product lines so you can inform them of an accessory for their product. At this moment, most CRM systems don’t yet include the AI technology to do this on their own, so humans have to take a lot of steps to build the list, create the template, send the email, respond, etc. etc. But very soon our CRM systems will catch up. So let’s assume that’s the case.
ChatGPT can take these directions and put together a campaign. But ChatGPT will need some love — or a better word: prompts — from the human. There will be questions about what customers to include and whether or not the messaging makes sense. ChatGPT will ask for help designing the template, scheduling the email and determining a subject line. ChatGPT will need to understand from a human how to tackle responses to the campaign. ChatGPT will also need to be prompted by a human to further analyze the data or perform more follow-up steps to the initial campaign.
ChatGPT is a conversational chatbot. It will need some back and forth to do its job.
AutoGPT does not need these prompts. If trained correctly it will make these decisions on its own. It will learn — by itself — how to reply to customer responses based on other responses. AutoGPT will, based on prior campaigns, make its own determinations for the email design, scheduling, graphics and subject line. It will choose on its own the targets of a campaign and then determine whether or not opens, views, clicks and responses are worth reporting back to management. It will independently create new lists of target prospects and customers based on who has or hasn’t responded to prior campaigns and then automatically reach out to those targets again based on past communications. It will literally perform the role of a marketing professional with minimal supervision. More actions. Less conversations.
And that’s just one use of AutoGPT in a CRM system. There will be many others, from behaving like an independent customer service representative answering questions on its own based on prior responses and other data, to following up on open orders and quotations in lieu of a salesperson. And, as technology expert Greg Isenberg predicts, it could help create a social media presence that “will understand every nuance of your community and create content & memes that have the highest probability of landing” and that will “change course based on data.”
AutoGPT does not need human prompts like ChatGPT. It is autonomous and independent. It’s also experimental, open source (you can get it on GitHub), requires programming (for now) and currently faces many challenges from accuracy to bias.
“AutoGPT has internet access, long-term and short-term memory management, text generation, and integration with 11 Labs (AI text to speech gen),” Isenberg writes. “And it operates AUTOMATICALLY without YOU.”
Yes, it’s scary — even terrifying — to some. But it’s coming. And it’s going to elevate the importance of your CRM system. Because for all of this to work, AI needs data and that data is in your CRM database.