OpenAI with a ton of announcements, on the heels of last week’s OpenAI feature called Instant Checkout, allowing users of ChatGPT to purchase items directly through the chat interface.
OpenAI has announced the general availability of Codex, a powerful coding assistant, along with three new features designed to enhance its utility for engineering teams. The introduction of a Slack integration allows users to delegate tasks and ask questions directly within their team channels, while the Codex Software Development Kit enables developers to embed its capabilities into their own workflows with minimal setup. Since its research preview launch in May, Codex has seen a significant increase in daily usage, growing more than tenfold since August. It now serves over 40 trillion tokens and is utilized by a range of companies, from startups like Duolingo to large enterprises such as Cisco and Rakuten. Cisco engineers report that Codex has cut their code review times by up to 50%, allowing them to focus more on transformative work instead of manual checks.
OpenAI has announced a new “Apps SDK” that enables third-party applications to launch and operate directly within ChatGPT, allowing users to interact with services like Zillow, Canva, and Spotify without leaving the chat interface. The Apps SDK supports features like inline app integration, fullscreen views, and picture-in-picture mode, enhancing user interaction while maintaining a conversational flow. OpenAI has emphasized strict privacy and safety standards for any apps integrated into ChatGPT.
OpenAI has announced a new feature that allows developers to create applications that function directly within ChatGPT, enhancing the user experience by enabling seamless integration with various software and services. This includes popular platforms such as Canva, Booking.com, and Zillow, which users can access to complete tasks while receiving contextual support from ChatGPT. During a recent demonstration, an OpenAI employee showcased how ChatGPT could interact with Canva to generate a business poster and subsequently create a pitch deck based on that design. As part of its broader integration strategy, OpenAI is set to expand this capability with additional apps in the coming weeks, including DoorDash and Uber. Developers can begin using the Software Development Kit for app creation immediately, with plans for a formal app submission process later this year.
OpenAI introduced significant updates to its application programming interface, unveiling the GPT-5 Pro language model and a new video generation model named Sora 2. These updates are part of a broader initiative to attract developers, including the launch of tools for building AI agents and applications within the ChatGPT environment. OpenAI highlighted that the GPT-5 Pro model is tailored for sectors such as finance, legal, and healthcare, where high accuracy and deep reasoning are essential. The new voice model, “gpt-realtime mini,” promises to provide low-latency interactions at a cost that is 70% lower than its predecessor, while maintaining comparable voice quality. Additionally, Sora 2 allows users to generate realistic audio and video content, enhancing creative control and synchronization of sound and visuals, making it particularly useful for concept development in various industries.
And in deals, OpenAI has secured a multibillion-dollar chip deal with Advanced Micro Devices, which includes an option to purchase a 10% stake in the chipmaker. This agreement, announced recently, is expected to significantly enhance OpenAI’s computing capabilities, crucial for advancing its artificial intelligence technologies. The deal involves the deployment of hundreds of thousands of AMD’s artificial intelligence chips over several years, beginning in the second half of 2026, with projections of generating tens of billions of dollars in annual revenue for AMD. Following the announcement, AMD’s shares surged more than 30%, adding approximately $80 billion to its market capitalization.
Why do we care?
OpenAI just dropped a truckload of announcements — new models, an app store, instant checkout, Slack integration, and even a deal to buy part of AMD.
Put it all together, and here’s what’s happening: OpenAI is building the Windows of AI. ChatGPT’s becoming the platform — not just the tool. You’ll buy, build, and run stuff right inside it.
And this isn’t happening in a vacuum — Microsoft’s Copilot ecosystem and Google’s Gemini suite are racing to do the same. The platform war for where work happens has officially started, and service providers will need to pick their lanes carefully. If you’re building client automation or workflow systems, this is where those integrations need to land — inside the AI workspace, not outside it.
That’s huge — and dangerous. Because if ChatGPT becomes the new desktop, you’re not the platform anymore. OpenAI is. MSPs that just resell or support tools will get squeezed out as users do everything inside the chat window.
So here’s what to do: start building. Use their SDKs, figure out how to automate client workflows, and be the expert on integrating this new layer — with governance, not just code.
We’ve seen this before: Windows owned the PC era, and iOS arguably owned mobile. If ChatGPT owns AI, service providers better learn how to play on that platform — or risk getting locked out entirely.