As OpenAI gears up for its first-ever developer conference, OpenAI DevDay, starting Monday, the anticipation is palpable. In the past 12 months, OpenAI has gone from virtual unknown to market dominance. And the AI juggernaut has been on a roll recently, releasing a flurry of acquisitions, innovative products and features over the summer and fall. It’s also had some significant setbacks. What are developers going to see on November 6? Here are some predictions, some speculation, and a wishlist.
This is the company’s first developer conference, which in many ways it represents the end of Act I for the little app that took over the world in twelve months. With that stunning a debut, expectations are sky high.
And has largely delivered. The corporate impact is massive. Imagine one day being given a serendipitous little gift that helps you run your business not just more efficiently, but much more effectively and with better experiences.
The ability for a sales rep to go into their CRM and access customer data in multiple ways that the rep themself can design – as someone who marketed CRM software for years, I know the frustration this will address, a frustration we are barely aware of because it’s true of every platform.
AI changes and opens up the data UX completely. ChatGPT and generative AI could save ERP companies like SAP billions, just by making their own software easier to use. It’s happening across the industry, from Cisco to ServiceNow: generative AI applications have multiple use cases in the enterprise, are being widely deployed, and are delivering exceptional experiences.
So interest in DevDay is very, very high. Most product announcements have been focused on end users; from the introduction of GPT-4V(ision) to the unveiling of a user-friendly UI for fine-tuning, the company has been pushing out an impressive set of integrations and expansions. What’s the next act? Expectations are high. Extremely high? How will DevDay’s announcements live up to Act One? What will OpenAI break our brains with next?
Whatever it is, it may not happen Monday.
- Expectations are being managed. A flurry of recent announcements, including GPT-4V(ision), a UI for fine-tuning, the Turbo Instinct release and the reactivation of ChatGPT’s “browse with Bing” feature, while impressive, doesn’t look like it will have many additions at DevDay.
- Despite the ongoing presence of Johny Ive, insiders are clear hardware is still a ways off.
- Also a no-go: Sam Altman, OpenAI’s CEO, has made it clear that there will be no announcements regarding GPT-5. This likely has something to focusing the abrupt scuttling of the Arrakis operating system just weeks before devs presumably would have seen it for the first time.
What Will Be Announced?
So what will be announced? Here’s a summary of the speculation we’ve heard around the developer community:
- Developers have been advocating for cost reductions on GPT-4, including the implementation of a session-based API to alleviate the burden of sending all messages in the current stateless API. One possibility is that OpenAI could upgrade all its models to GPT-4 or beyond, making GPT-3.5 open source, which would be a significant announcement and disrupt the open-source model landscape.
- OpenAI is reportedly venturing into the hardware business. Partnering with WHOOP to launch WHOOP Coach, a generative AI feature for wearables, suggests OpenAI may integrate its AI capabilities into various hardware designs. Discussions between Jony Ive, the iPhone’s renowned designer, and Sam Altman indicate a potential AI hardware project. While competing in the smartphone market is unlikely, OpenAI may focus on simpler hardware applications like smartwatches or smart rings.
- OpenAI could also explore opportunities in XR headsets or computer vision products, given its development of GPT-4V(ision). Recent acquisitions, such as Global Illumination, a 3D world designing company, hint at OpenAI’s interest in a metaverse-style simulation with AI bots, potentially leading to a move into the gaming industry.
- Significant updates to its developer tools and capabilities to make it more cost-effective and versatile for developers to build software applications based on its AI models.
- Addition of memory storage to developer tools, which could reduce costs for application makers substantially, addressing concerns about the expenses associated with using OpenAI’s powerful models.
- Introduction of vision capabilities, enabling developers to create applications capable of analyzing and describing images. This feature has potential applications in fields ranging from entertainment to medicine.
These updates reflect OpenAI’s and Altman’s ambition to evolve from a consumer sensation into a thriving developer platform and attract more companies to utilize its technology.
Major Updates To Developer Tools
OpenAI is making significant updates to its developer tools and capabilities to reduce costs and attract more developers and companies to utilize its AI models. These updates, including memory storage and vision capabilities, aim to make OpenAI indispensable in various industries. Let’s see how this strategic vision is to become a key player in AI development, addressing challenges and competition in the rapidly growing AI startup landscape, is received.
OpenAI will officially unveil these new features on Monday, with the goal of expanding corporate deployments that are leveraging OpenAI’s technology to build AI-powered chatbots. The next evolution, autonomous agents that can perform tasks without human intervention, may also be announced, addressing a key strategic objective for the company.
OpenAI’s strategic vision is to become indispensable to other companies building applications, ranging from DoorDash to writing assistants like Jasper. This expansion is crucial as OpenAI aims to sustain its rapid growth. The company expects to close this year with $200 million in revenue and has set its sights on reaching $1 billion by 2024.
However, OpenAI has faced challenges in enticing external developers and companies to build businesses using its technology. To overcome these challenges, the company plans to release some API-based capabilities:
- A “stateful API” (Application Program Interface), will allow companies to create applications that remember the conversation history of inquiries. This can significantly reduce usage costs for developers. Currently, processing a one-page document using GPT-4 could cost 10 cents, depending on the complexity of the input and output.
- A vision API, enabling developers to build software that can analyze images. This move marks a significant step for OpenAI in offering “multi-modal capabilities,” allowing the processing and generation of different types of media, including text, images, audio, and video.
These updates are designed to attract more developers to pay for access to OpenAI’s models, empowering them to create their AI software for various purposes, such as writing assistants or customer service bots. Investors have poured over $20 billion into AI startups this year, many of which rely on OpenAI or other foundation model companies’ technology. However, investors are concerned about the vulnerability of these startups to replication by rivals or larger companies through product updates.
OpenAI faces competition not only from other foundation model companies but also from startups experimenting with alternative AI models and open-source options, like Meta’s Llama. To distinguish itself from deep-pocketed monster rivals like Google, OpenAI must keep developers satisfied.
Wishlist + Areas for Improvement
OpenAI needs to shore up its core product and fix plugins, which have not been well received. OpenAI previously released ChatGPT plugins, add-on tools allowing developers to create applications within ChatGPT. The company hoped these plugins would become equivalent to Apple’s iOS App Store, providing a competitive edge. While some plugins initially garnered hype, their popularity waned over time. For instance, the Scholar AI plugin had approximately 7,000 daily users as of late August, compared to ChatGPT’s 180 million monthly active users.
Altman has acknowledged the need for improvement in this area. He admitted that the plugins did not gain the market traction they had hoped for. The company is now focusing on these new updates and features to attract and retain developers, making OpenAI’s technology an integral part of the AI development landscape.
On our wishlist are several possibilities with immediate potential and some much further out. Improved search capabilities, an AGI progress report, AI-ready hardware, any update on AI/robotics integration, major industry projects, and any updates on resource utilization and capabilities movement are high on our list of desired dev options. The prospect of enhanced animation and sound capabilities, as well as the introduction of snapshot summaries and multiple data use configurations, has also piqued a great deal of interest from the creative community.
Real-time Data Integration
Live, real-time external data source integration for ChatGPT is another area where OpenAI could make substantial strides, offering users seamless access to up-to-the-minute information. Recommendations and opinions driven by AI could revolutionize decision-making processes across various industries. Long term, it’s a Google search killer.
OpenAI’s AI team is also poised to make an impact with the potential launch of new training modules tailored to specific roles in new verticals. From bookkeepers and customer service representatives to accountants, project managers, project designers, and copywriters, these modules could empower individuals to harness the power of AI in their respective fields.
As the countdown to OpenAI DevDay continues, the tech world is agog with excitement and anticipation not seen for years. While GPT-5 may be off the table for now, the possibilities for innovation and advancement in the world of AI seem limitless. OpenAI’s mere existence pushes the boundaries of what’s possible. More on Monday!
Written with support from ChatGPT