When OpenAI unveiled ChatGPT Apps and its new Apps software development kit (SDK), it had the effect of a small bomb. ChatGPT entered the race for the West’s super-app, à la the Chinese WeChat app.

With more than 800 million weekly users, ChatGPT is no longer a B2B tool; it’s becoming a consumer platform. And, for publishers, this shift could transform how audiences find, consume, and interact with them.

From chat to platform: a post-search ecosystem

The new Apps SDK allows third-party services — from Spotify to Canva, Booking.comm or Coursera — to live inside ChatGPT. There’s no need to switch between tabs or apps. Users can now book a trip, play music, or design slides directly within the chat.

In practical terms, this marks the arrival of a WeChat-style Internet — a single conversational interface where services, transactions, and content coexist seamlessly. It’s interesting to note ChatGPT also announced the integration of more than 1 million Shopify vendors directly in the chat and launched the agentic commerce protocol (ACP) to support transactions.

For media companies, this means one thing: ChatGPT is becoming both a distribution layer and a user interface.

The end of clicks and the rise of “conversational discovery”

Traditional search engine optimization (SEO) and social discovery are being replaced by what researchers now call generative engine optimisation (GEO) and generative AI optimisation (GAIO). This is the art of making your content visible inside generative systems.

In this new landscape, zero-click consumption accelerates. Users won’t “visit” your site. Instead, they’ll ask ChatGPT a question, and your journalism might power the answer in the background.

This means:

• Traffic attribution will fade.
• Your brand might become invisible if it’s not properly structured, tagged, and licensed for AI access.
• Value shifts from audience volume to content accessibility, structure, and trust.

Risks and opportunities for media organisations

This change in the media landscape also highlights several risks and opportunities media companies should note.

Key risks include:

• Loss of direct traffic and traditional analytics visibility.
• Reduced brand exposure if AI assistants summarise without clear attribution.
• Complex data governance and consent management for AI access.

However, there is a huge opportunity to be a first mover in the ChatGPT ecosystem by building micro-apps, such as those focused around data, events, Q&As, news digests, live updates, and topic explainers.

Other opportunities include:

• Turning journalism into actionable experiences, that is from “read” to “do.”
• Distributing the data media companies own straight into their audiences’ conversations.
• Integrating new monetisation layers once OpenAI enables in-chat commerce and paid actions.

In short, the question is not whether media will integrate into AI platforms, but how intentionally they will do it.

Recommendations for media leaders

As the course for an AI-enabled future continues to be created in the media ecosystem, here are five recommendations for media leaders to consider:

1. Map use cases. Identify which user intents (e.g., explain, compare, analyse) could trigger your content in ChatGPT.
2. Structure your data. Implement schema, meta-data, and clear attribution to make content readable for AI.
3. Expose model context protocols (MCP), not just pages. Turn your content management system (CMS) into an MCP-ready backend that AI systems can query safely.
4. Prototype light apps. Test one conversational module before building full integrations.
5. Govern data and tone. Ensure editorial transparency, permission control, and an “AI-native” brand voice.

Looking ahead: from “media as destination” to “media as infrastructure”

Tomorrow’s audience won’t browse — they’ll converse. In this environment, publishers that see themselves not as content factories but as trusted data and insights providers will lead.

OpenAI’s move toward embedded apps signals a post-browser Internet, where the value of a media brand will depend on its ability to integrate, interoperate, and remain visible inside AI ecosystems.

The time to adapt isn’t in six months. It’s now.

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .