Time's almost up! There's only one week left to request an invite to The AI Impact Tour on June 5th. Don't miss out on this incredible opportunity to explore various methods for auditing AI models. Find out how you can attend here.
Among the many new features OpenAI CEO Sam Altman announced yesterday at the company’s first-ever developer conference, DevDay — the most important may have been the new GPT Builder.
This tool — rolling out slowly for ChatGPT Plus and ChatGPT for Enterprise subscribers — allows users to create their own GPTs, essentially AI agents atop OpenAI’s new GPT-4 Turbo model, using only plain English typed commands. This opens the door for anyone — even non technical users or those with zero formal developer training — to build their own AI agents and applications in a matter of minutes.
Such third-party GPTs can reference documents and materials uploaded by the user, and perform repeatable actions they have specified even accessing other apps, say, searching for calendar conflicts and automatically messaging other attendees to a meeting (one example from Zapier shown off on stage). OpenAI said that it will make third-party GPTs available in a GPT Store, and will share revenue it generates from their usage with the creators.
The GPT Builder is not widely available yet, but several users have gotten early access, and are reporting it is indeed easy, fast, and requires no prior coding knowledge nor developer training to be able to build third-party GPTs. Here are some examples of the early GPTs said users have built.
June 5th: The AI Audit in NYC
Join us next week in NYC to engage with top executive leaders, delving into strategies for auditing AI models to ensure optimal performance and accuracy across your organization. Secure your attendance for this exclusive invite-only event.
Product prototyping
Leveraging ChatGPT’s browsing capabilities with Microsoft Bing and its integration with OpenAI’s new DALL-E 3 image generator, University of Pennsylvania Wharton School of Business professor and AI influencer Ethan Mollick shared a video showing off his GPT called “Trend analyzer,” that looks up market trends in a particular segment and then creates prototype images of a new product for the user to design or pursue in real life.
Simpsonize Me GPT
Another new GPT that leverages the DALL-E 3 integration as well as ChatGPT’s new “All Tools” mode to reference an image uploaded by the user, Simponize Me GPT by Octane AI CEO Matt Schlicht automatically applies a prompt to turn a user’s uploaded profile photo into a cartoony image reminiscent of the style of Matt Groening’s long-running animated comedy series The Simpsons. Schlicht wrote that he built it in “under 10 minutes.”
Maximizing social engagement on X
AI influencer Rowan Cheung, creator of The Rundown AI newsletter, created X Optimizer GPT, which automatically analyzes his proposed text for posts from his account on the social network X (formerly Twitter) and suggests improvements and optimal times to post for maximum engagement from the network.
He posted that he built it “on the spot” by downloading his X/Twitter post data and uploading it to ChatGPT for analysis.
Making animated GIFs with Gif-PT
Leveraging DALL-E 3’s image generation capabilities, app developer and former Twitter employee Nick Dobos posted on his X account that he created a new GPT called “Gif-PT” that automatically applies the proper prompts to create multiple grids of images that are, in turn, turned into frames. Using ChatGPT’s Advanced Data Analysis mode, aka Code Interpreter, it writes Python code and converts the frames into a single animated GIF that the user can download.
While Dobos admitted the results can be inconsistent and “janky” they are impressive given the level of work put into the app, and the fact that it is analogous to Meta’s own AI animated sticker generator, but made by only one person in presumably a fraction of the time. He said he was “very impressed” in a subsequent post on X.
Coaching and mindfulness
The very first customizable GPT that OpenAI CEO Sam Altman showed to the world — by building it live on stage in about five minutes during his DevDay keynote address — was a coach for tech company founders based on his prior experiences receiving advice from VCs, and dispensing it.
So it makes sense that some of the earliest third-party GPTs created outside of OpenAI would also follow in this mold. Two examples include another product coach GPT by Yana Welinder, CEO and founder of product copilot company Kraftful, and a daily zen guide GPT by Mustafa Ergisi, founder of ai2sql.
While Welinder’s provides practical strategic business advice such as how to improve retention and pull together case studies of sample products, Ergisi’s provides mindfulness exercises and suggestions of habits to produce better sleep.
Of course, these are just the very first few GPTs developed in the day since OpenAI announced its new GPT Builder. There will be many more to come, presumably with many more features and capabilities than these initial ones.
But, as someone who lived through the initial wave of the Apple App Store and all the silly simulated beer drinking and fart noise and lightsaber apps, this initial wave of GPTs from third-parties is a strong start for GPT Builder and the GPT Store, and OpenAI’s ambitions to be to AI what Apple was to mobile.
