Inside OpenAI’s DevDay 2025: Apps, Agents, and a Smarter GPT-5
AI just took another leap forward for both developers and anyone shaping customer experiences. At OpenAI’s DevDay 2025, Sam Altman introduced a lineup that turns ChatGPT from a productivity tool into a true platform. The new Apps SDK, Agent Kit, and GPT-5 Pro API are designed to make AI apps feel as natural as a conversation.
Altman summed it up best: “AI has moved from systems you can ask anything, to systems you can ask to do anything.” For marketers and creative technologists, that shift opens a new channel, one where discovery, engagement, and conversion all happen inside the same chat.
From Chat to Platform
The scale behind this launch is staggering. Since DevDay 2023, OpenAI’s developer community has doubled to four million, and ChatGPT now reaches over 800 million weekly users. That user base effectively makes ChatGPT its own digital ecosystem. It’s part search engine, part app store, and part assistant.
Instead of connecting external tools through plug-ins, brands can now build native apps inside ChatGPT. This is where the new Apps SDK comes in.
The Apps SDK: Turning Chat Into a New Marketing Channel
The Apps SDK lets developers (and soon, creative teams) design interactive experiences that live directly within ChatGPT. Built on an open standard called MCP, it gives full control over backend data, triggers, and user interface, all without forcing the user to leave the conversation.
In live demos, Canva generated an entire pitch deck from a single chat prompt, Zillow embedded an interactive home search map, and Coursera streamed a video that ChatGPT could answer questions about in real time. These aren’t gimmicks, they’re early examples of what conversational user experience looks like when it’s both functional and context-aware.
For marketers, this means you could soon launch in-chat product finders, style guides, or learning tools that feel personal and frictionless. Apps can be discovered by name (“Find this app for me”) or through recommendations surfaced mid-conversation; a new kind of organic reach powered by intent. Monetization features are also coming soon, making ChatGPT a distribution channel in its own right.
Connecting brand data directly into ChatGPT is powerful, and before doing so, the user will be asked to give permission, but there should still be a question of who owns that data. When agents access customer information or internal systems, marketers face a compliance gray zone. How is that data stored? Can OpenAI models learn from it? Powerful, useful, necessary for the future, yes, but awareness is key.
Agent Kit: Automating Brand Expertise
If the Apps SDK is about building experiences, Agent Kit is about scaling intelligence. It’s a visual toolkit that lets teams design, test, and deploy their own AI agents without heavy code or infrastructure.
The Agent Builder provides a drag-and-drop canvas for workflows; the Chat Kit lets you embed branded UIs into your site or product, and the Evals for Agents system helps you measure performance and safety.
Real-world examples are already emerging. HubSpot used Agent Kit to upgrade its assistant “Breeze,” which now searches policy documents and customer data in real time, and Albertsons built an internal agent that helps store managers analyze sales trends.
For marketers, the implications are clear: you can now create an assistant that actually knows your product line, tone, and audience; a true brand expert that lives inside the same channels your customers already use.
Voice, Video, and the Human Layer
Two final updates matter for content creators. The GPT-Realtime-Mini voice model now offers near-studio quality at 70% lower cost, which could power more natural voice search, narrated ads, or branded audio experiences. And the new Sora 2 video model, available in preview, brings controllable motion and synchronized sound.
In one showcase, Mattel used Sora 2 to turn toy sketches into animated concepts within minutes. For digital marketers, that’s the future of campaign creative: producing social-ready, sound-synced visuals directly from text or product imagery.
Platform Centralization
ChatGPT’s new role as an app platform gives OpenAI massive distribution power, and equally massive control. Much like Apple App Store, it decides which apps are seen, promoted, and monetized. That’s great for quality assurance but risky for competition. If marketers begin relying on ChatGPT for discovery or conversion, they’ll effectively be building on someone else’s property. The upside is reach, the downside is dependence.
What This Means for You
OpenAI’s message this year was simple: it’s never been faster to go from idea to product. For marketers, that means it’s never been easier to turn audience insight into interactive experience.
Start by mapping your customer journey. Identify where a conversational app could simplify decision-making or where an agent could personalize follow-ups. Experiment with a single use case (a guided product quiz, a digital concierge, a learning bot) and build from there.
As these tools evolve, the brands that learn to design for chat-first experiences will define the next generation of marketing. The future isn’t about adding more channels, it’s about making your brand do more inside the channels customers already use.
