ML6 • Blog

Rethinking the customer journey with ChatGPT apps

Written by Julian Zabbarov | Feb 17, 2026 8:58:16 AM

Executive Summary

The way users discover, evaluate, and interact with digital services is changing. As conversational AI tools like ChatGPT become a primary entry point to information, traditional search-driven customer journeys are increasingly fragmented. Users move between platforms, repeat context, and switch interfaces to complete even simple tasks.

ChatGPT apps introduce a different model. Services can be invoked directly within a conversation and—crucially—interactive user interfaces are rendered inside the chat itself. This collapses discovery, decision-making, and action into a single, continuous flow. Instead of navigating between chatbots, websites, and apps, users can remain within ChatGPT to complete tasks such as booking holidays or shopping online.

This shift has important implications for customer experience, product discovery, and brand visibility. Companies must think beyond websites and search rankings, and design services that can operate effectively inside AI-mediated interfaces. This article explores how ChatGPT apps reshape the customer journey, how they work architecturally, and what this means for organizations preparing for a future where conversational interfaces become a key distribution layer.

From search to chat

For years, online visibility followed a well-understood playbook. If you wanted to be found, you optimized for search engines. You invested in SEO, climbed Google’s rankings, and guided users from search results to carefully designed landing pages.

That playbook is no longer sufficient. Today, an increasing number of users turn directly to chatbots such as ChatGPT, Gemini, or Claude to find answers. In a recent study by Business Harvard Review, 65% of young adults said to use chatbots as a direct replacement for Google search. The transition from search to chat continues to accelerate: Gartner predicts that traditional search engine volume will decline by 25% by 2026 due to AI chatbots and virtual agents, while industry reporting (including Reuters) indicates that ChatGPT alone now reaches more than 800 million weekly active users.

 

It’s still a visibility game

Understandably, this shift creates concern for companies. If users stop primarily relying on search engines, how do you remain visible? How do you build awareness around your brand and generate user traffic? Traditional SEO and even newer concepts like Generative Engine Optimization are no longer enough as conversational AI becomes a primary gateway to information and services. As Gartner summarizes the challenge: “Winning visibility now means optimizing for both AI-driven answers and classic search results, with content that is specific, conversational, and trustworthy.”

Gaining user traction with ChatGPT apps

 

Apps In Chat GPT

OpenAI recently announced a feature that could help address this challenge: ChatGPT apps. ChatGPT apps allow companies to expose their services directly within a ChatGPT conversation, making them accessible at the moment they are relevant.

A ChatGPT app can be invoked by referencing the service after an “@”. Depending on the user’s request, the app dynamically decides to respond. The interaction happens through a user interface rendered directly in the chat response, allowing users to engage with the service without leaving the conversation.

To make the concept more tangible let’s take a look at an example: Researching and booking hotels.

 

The customer journey so far

Before ChatGPT apps, a typical customer journey would look like the following: The user might start by asking ChatGPT for travel inspiration. Based on the user's preferences, like budget, weather preferences, or hobbies, ChatGPT and the user interactively come up with a travel itinerary. Once the user decided on a travel plan, he or she would switch context from ChatGPT to other platforms to book the flights and hotels. The user has to search again, filter options, and complete a reservation.

A customer journey with ChatGPT apps

With ChatGPT apps, this fragmentation disappears. The user still gets travel recommendations from ChatGPT. However, when it comes time to choose an accommodation or select flights, ChatGPT can invoke the booking.com app mid-conversation so that the entire customer journey unfolds within a single conversation.

An integrated ChatGPT app has the following capabilities:

  • Know: The app can filter relevant options because ChatGPT naturally passes on relevant information when invoking the app.
  • Show: The user can interact with the service using a UI component that is displayed in the ChatGPT response.
  • Do: The user can take action directly in the app. With an integrated payment functionality it is possible to complete the booking process without ever leaving ChatGPT.

This naturally raises the question of how users can explore the available services in ChatGPT. The first option is to browse the ChatGPT app store, similar to the app store we know from our phones. A second, plausible future flow could involve sponsored placements appearing directly within conversational responses, allowing users to invoke an app from an ad. For companies, this reinforces the importance of not only building ChatGPT apps, but also preparing for a world where conversational discovery combines organic presence through apps with paid visibility.

 

From point-and-click to natural language

This is a point worth pausing on. Displaying and invoking apps mid-conversation does more than create new opportunities for brand visibility. It represents a shift in how we will interact with digital services. They move us away from static, point-and-click interfaces toward natural language as the primary interaction layer for searching, discovering, and shopping online - all within a single, continuous context.

In this setting, conversations become the interaction layer between users and services. Instead of navigating websites, users now express goals and preferences in conversations, allowing intent to be captured early and with greater precision. Discovery, decision-making, and action increasingly collapse into one adaptive flow, where services are invoked when they are actually relevant. The result is a more personalized and intuitive customer journey in which a user interface is only shown when it adds to the user experience.

 

Building the customer journey of the future

At ML6, we aim to “walk the talk.” To explore what customer journeys with ChatGPT apps could look like in practice, we experimented with them ourselves. At a recent hackathon with OpenAI, we built a ChatGPT app that rethinks how users shop at a furniture store.

This is IDEA, the future way of shopping at a furniture store. With the IDEA app, users can browse a catalog, select items, and add them to a cart using natural language. They can also ask detailed questions about products—such as materials, dimensions, or compatibility—without interrupting the flow.

To go beyond traditional e-commerce experiences, we extended the app with vision capabilities. Users can visualize selected furniture items directly in their own rooms, adjust placement using natural language, and iterate quickly on different configurations. Instead of imagining how a product might look, users can explore it interactively, gaining confidence before committing to a purchase.

 

An architectural view on ChatGPT Apps

Under the hood, ChatGPT apps rely on a clear and modular architecture. At a high level, a basic app consists of three core components.

  • ChatGPT: The main conversation with the user is done in ChatGPT. From a developer perspective, the main step is registering the app by providing a name, description, icon, and a link to the MCP server. Based on the app description, ChatGPT can decide when it is appropriate to invoke the app during a conversation.
  • MCP Server: The MCP server contains the application logic. The MCP server exposes a collection of tools that ChatGPT can call. Based on the tool names and descriptions, ChatGPT selects the appropriate tool at runtime. Each tool handles a specific task. It can connect to customer APIs to fetch relevant information such as product catalogs or images. The server returns structured data, as well as HTML snippets that reference hosted JavaScript and CSS components.
  • UI assets: The UI assets are responsible for rendering the experience and making the experience interactive. Each tool has an associated UI component that visualizes the structured data returned by the MCP server. The HTML references the hosted assets to ensure a consistent and interactive presentation inside ChatGPT.

The app we presented at the OpenAI hackathon extends this architecture further. To visualize selected items in a user’s room, we developed a tool that retrieves the current state of the shopping cart including the number of selected items. We upload relevant product images to the Files API, and references those files in a custom prompt sent to the OpenAI Images API.



Interactive chatbot apps came to stay

What started with ChatGPT apps is already maturing into an industry-wide standard through the existing Model Context Protocol (MCP). MCP is an open protocol that defines how language models connect to external tools and services, enabling AI systems to move beyond text generation and interact with real-world functionality in a structured and interoperable way.

In January 2026, Anthropic introduced MCP Apps as an official extension to the protocol. Anthropic created and will continue to develop MCP apps in a joint initiative with OpenAI to establish an industry standard for how tools defined in an MCP server can reference user interface assets to render structured data directly inside conversations.

This allows companies to build interactive, UI-driven chatbot apps once and integrate them across multiple platforms. An app architecture designed for ChatGPT can also be reused in other assistants, such as Anthropic’s Claude. As a result, the MCP Apps standard reduces platform lock-in and positions conversational apps as a portable, future-proof channel for delivering digital services.

ChatGPT – The landing page of the future?

In the meantime, the adoption of ChatGPT apps is undergoing. With the integration of major services like Alltrails, OpenTable and Spotify, users can find hikes, book restaurants and create custom playlists in ChatGPT already today.

Through the integration of apps, ChatGPT is evolving from a standalone chatbot into a platform orchestrator. It increasingly mediates how users discover and interact with digital services, positioning ChatGPT as the primary interface between users and providers.

For companies, this changes the competitive landscape. Success is no longer determined solely by owning a website that’s present on search engines, but by how well your services help users achieve their goals in ChatGPT conversations. Visibility and relevance are increasingly negotiated at the platform level, making it essential to think beyond traditional distribution channels.

 

A window of opportunity

The ongoing shift from search to chat creates a limited window of opportunity. With ChatGPT apps still in beta and the ecosystem in its early stages, companies that move now can build a leading position on the ChatGPT app store and strengthen the brand perception as an innovator. Particularly younger, digitally native generations are heavily relying on chatbots for everyday tasks (Harvard Business Review). Being among the first to offer a seamless service experience inside ChatGPT presents a chance to secure visibility amongst those that count on chatbots to navigate the web and their daily lives.

If you are considering how to bring your service into ChatGPT, ML6 is your partner. We help organizations design, build, and scale AI-powered customer journeys that are ready for what comes next.