What AI-Curious Brands Need to Know from OpenAI’s Inaugural Developer Conference

Editor's note: This is an abridged version of an article originally published on the IPG Media Lab blog. If you wish to get the latest insights from us, please consider following us on Medium and Twitter @ipglab.

On Monday, OpenAI, the company behind ChatGPT, hosted its first-ever developer conference in San Francisco. Kicking off the opening keynote, which had a tight 45-minute runtime, CEO Sam Altman quickly shared a slew of impressive stats demonstrating the popularity of its products — ChatGPT now has 100 million weekly active users, 2 million developers are now working with OpenAI’s tools, and around 460 companies on the Fortune 500 companies are using its products — before jumping right into the updates coming to the AI chatbot that kicked off this current wave of AI frenzy.?

Among all the updates OpenAI shared on Monday, of which you can find a comprehensive list on The Verge, including a “GPT-4 Turbo” model that boasts a remarkable 128,000 context window that enables it to process the equivalent of more than 300 pages of text in a single prompt, three announcements stood out as must-know for brands looking to leverage the potential power of generative AI in their marketing and beyond.

Time to Build Branded GPT Bots

One of the key things that OpenAI announced at this developer event is the introduction of custom chatbots called GPTs, based on ChatGPT’s underlying large language models (LLMs) along with a new GPT Builder that will allow anyone to customize and share their own AI assistants with natural language.?

GPTs are new custom models meant to be tailored to specific uses. Through the new Builder interface, you’ll be able to add custom instructions, knowledge, and actions, simply by typing what you want them to do. For example, Altman live demoed the process of building a chatbot that specializes in giving feedback to startup founders.?

This marks the latest evolution of Custom Instructions, a feature that OpenAI introduced in July to let users set some preferences that they’d like ChatGPT to consider when generating its responses. Now, all of these GPT agents can be created and accessed as a standalone chatbot, which helps improve discoverability and user-friendliness.?

For brands, the arrival of GPTs is a long time coming. Ever since ChatGPT blew up, brands have been wrestling with the question of how to deliver utility-driven customer value through AI chatbots. Now with GPTs, brands can tap into OpenAI’s industry-leading LLMs to create custom chatbots that are tailored to their specific brand voice and identity. The ability to add custom instructions and actions will enable brands to create chatbots that excel in specific domains by feeding them their first-party data and insights. For example, a retailer can build a GPT bot for holiday gift shopping that acts as a virtual shopping assistant, helping customers find the best gifts available in their inventory and delivering recommendations in a style that best suits their brand identity.??

It is interesting to contrast GPTs with what Meta recently announced in its AI offerings for brands. Not only can advertisers now use embedded generative AI tools in Meta’s Ads Manager to create background images and write ad copies, they may also soon get a chance to build AI Personas that are accessible across Meta’s suite of messaging and social platforms.? Considering most brands have established a presence on Meta’s platform, OpenAI still has a lot of catching-up to do if it wants to convince AI-curious brands that its budding platform is worth their investment.?

And in case some brands still have cold feet when it comes to experimenting with AI due to legal concerns, OpenAI wants to ease those anxieties with a new “copyright shield” feature that helps defend its customers from legal action, going as far as promising to pay the costs incurred if you face legal claims around copyright infringement.

Bottom line: brands can use GPTs to deliver superior customer value and build stronger relationships with their customers by creating personalized, helpful, and engaging chatbots. And the development cost is also significantly reduced thanks to the new Builder tool, which requires no coding knowledge. In addition, this obviously lowers the barrier to entry and allows small and medium-sized businesses, who don’t necessarily have the resources for building custom AI chatbots, to be on a leveled playing field.?

Gearing Up for the “App Store” of AI Chatbots?

Of course, the GPT Builder tool is not only limited to brands and enterprise users. Any regular developer can access it to create their own custom chatbots. In fact, OpenAI is counting on an explosion of custom chatbots to generate some much-needed revenue. OpenAI unveiled plans to open a GPT Store later this month where users can share and sell their custom GPT bots, with the store essentially functioning as a aggregator platform for users to discover and purchase.

This is a move that recalls Apple’s decision to launch the App Store for developers to sell their apps, which kicked off the mobile economy as we know it. But there’s no guarantee that this new GPT store will become the next App Store. Ironically, all of the issues that plagued the early days of the App Store will likely become headaches that OpenAI will have to deal with soon.?

First of all, a robust revenue-sharing program will need to be established. While Atman promised that the company will pay “people who build the most useful and the most used GPTs” a portion of the company’s revenue, no specific details about how that will be calculated or split between OpenAI and individual developers has been made available at this time. If OpenAI is serious about being an aggregator platform that facilitates the extension of the creator economy into the chatbot realm, it needs to develop a fair and transparent revenue-sharing program that earns developer and customer trust.

Secondly, OpenAI will need to address the issue of copyright and potential IP theft. Beyond the aforementioned “copyright shield” feature, OpenAI will need to establish a robust set of rules that ensures that developers have the right to sell their custom GPTs to prevent bad agents from plagiarizing or stealing GPTs from brands and other developers.?

Ultimately, the success of this GPT store will hinge on the quality and utility of the custom chatbots created by developers. Unlike the early days of the App Store, where there was a barrier to entry of app-coding knowledge for app development, the new conversational GPT builder tool will allow even more people to create custom GPTs and potentially cash in on this AI gold rush (especially if OpenAI’s revenue-sharing split turns out to be generous). Therefore, it is important that OpenAI develops a review system to ensure that the bots available in the GPT store are genuinely useful and free from potential biases.

For brands, the launch of a GPT Store would create a new battleground for brands to size up the competition and fight for consumer attention. But just like the mad dash to create mobile app turned out to be a fool’s errand for most companies — the App Stores quickly became saturated with millions of apps, making it difficult for new apps to stand out without significant marketing budgets, and most branded apps didn't offer a distinct advantage over existing solutions or web-based alternatives — the same could happen with the GPT Store and branded AI chatbots. That said, there is undeniable value in being among the firsts to establish a presence on this nascent platform and earning organic press coverage.

Building the Next-Gen In-App Assistants?

For brands that are focused on improving their own mobile experience and curious about how generative AI can help enhance the customer experience, OpenAI announced a new Assistants API, which lets developers add “assistants” to their apps that can call on OpenAI’s generative AI models and tools to perform tasks. Thanks to its partnership with Microsoft, the costs for using those APIs to build AI-driven apps have also been reduced.?

Previous to this developer event, some innovation-forward brands had already tapped into OpenAI’s existing APIs to build in-app functions. For example, Instacart launched in May a “Ask Instacart” search tool in its app, which is powered by OpenAI’s LLMs and designed to help customers save time on grocery shopping by offering personalized recommendations. Other companies that have tapped OpenAI to launch in-app chatbots include Expedia and OpenTable.?

Compared to the previous wave of in-app chatbots powered by OpenAI’s API, the new Assistants API offers a turn-key solution that makes it even easier for brands to build AI assistants and incorporate them into their apps. By using the Assistants API, brands can enhance their app design and make it more conversational and intuitive for users.

In addition, OpenAI also introduced Custom Models, which allows enterprise users to create a privately hosted LLM with proprietary and custom data, so that they are better at handling business-specific use cases, and allows companies to avoid security concerns regarding sending data to OpenAI's API. This would cost around $2 million to $3 million, so most companies are unlikely to have the budget for it, thus leaving the Assistants API as a more scalable solution for companies.??

What’s Next??

Needless to say, competition in the generative AI space is fierce, with several tech giants and startups already offering their own solutions. It will be fascinating to see how Google and other AI startups respond to these major announcements from OpenAI in the coming months. In order to succeed, OpenAI will need to provide a compelling value proposition to attract both developers and customers to its GPT store.?

For brand marketers, it is best to explore this uncharted territory in customer engagement and innovation with caution. Through the utilization of custom chatbots, brands may foster deeper connections with customers, and carve out a distinctive presence through AI-driven consumer interactions.