The App Store isn't the only way people will find your app

The traditional app discovery journey is relatively straightforward. You have a need, so you search for an app that might solve it. You might search for “running tracker”, “budgeting app”, “train tickets”, “meal planner” or “photo editor”. You are effectively translating your problem into the language of the app marketplace.

The app store then gives you a collection of products that it believes match that search. The problem is that people don't always think in product categories. They think about what they are trying to do.

Someone doesn't necessarily wake up thinking, “I need a personal finance application with budgeting functionality.” They might think, “I need to work out where all my money is going this month.”

Someone travelling might not think, “I need a rail booking application.” They might think, “I need to get from Leeds to London tomorrow morning and I'd rather not change trains.”

Someone managing a field team doesn't necessarily think, “I need a workforce management platform.” They might think, “I need to know which engineer can get to this job today.”

This is where conversational AI changes the starting point. Instead of translating the problem into something that a search engine or app store understands, the person can describe the problem in their own words and let the system interpret the intent behind it. That is a much more natural interaction.

It also creates a new question for product teams: if someone describes the problem your product solves without mentioning your product category, will the system understand that your app can help? That's a very different kind of discoverability.

It is easy to think of this as something that will happen several years from now, but people are already using generative AI to research products, services and decisions.

Adobe's analysis of more than a trillion visits to US retail websites found that traffic arriving from generative AI sources increased by 1,200% in February 2025 compared with July 2024. In the same research, 39% of surveyed US consumers said they had used generative AI for online shopping, with product research and recommendations among the most common uses.

The numbers have continued moving in the same direction. Adobe reported that generative AI referrals to retail websites were up 693.4% year-on-year during the 2025 holiday shopping season, while those AI-referred visits converted 31% better than other traffic sources.

That doesn't mean everyone has suddenly abandoned Google or started shopping entirely through AI. The proportion of traffic coming from AI remains smaller than established channels, and these figures are primarily US data. What the numbers do show is something more useful: AI is becoming another layer between a person's need and the product or service they eventually choose. And that layer is becoming increasingly capable of doing more than simply providing a link.

Recommending an app is one thing, but being able to actually use it is another. Imagine saying: “Find me a train to London tomorrow morning, book the cheapest flexible option and put it in my calendar.” In a traditional journey, you would probably move between several interfaces. You might search for trains, compare options, choose a ticket, enter your details, complete the payment and then add the journey to your calendar.

An AI assistant could eventually become the layer coordinating that experience. It could understand what you're asking for, identify the services capable of completing each part of the task, pass the relevant information between them and ask for confirmation where necessary.

The important thing here is that the user isn't necessarily thinking about individual apps. They're thinking about the task.

Apple is already building towards this model with App Intents, which allows developers to make an app's actions and content discoverable by Apple Intelligence and available through experiences such as Siri, Spotlight and Shortcuts. Apple's current developer documentation describes App Intents as the mechanism for connecting an app's capabilities to Apple Intelligence and Siri AI.

Apple's 2026 documentation goes further, describing how app actions can be understood through structured schemas so that Siri can interpret natural language and connect it to the appropriate action. Apple also describes experiences where actions can work across apps, allowing a person to move content or complete parts of a task between different services.

That is an important change in the role of an app. The app is no longer necessarily the place where the entire interaction has to happen. It can become a capability that another interface calls upon.

This is probably the biggest idea for product owners to get their heads around. For a long time, the goal has been to get someone into your app. You want them to download it, open it, navigate through it, complete a task and come back again.

But AI assistants introduce another possibility. Your product could provide a particular capability that someone accesses without consciously thinking about opening your app. Think about a hotel booking app… Its value isn't really “having a mobile application”, its value is helping someone find and book somewhere to stay.

• A travel app might provide live journey information.
• A finance app might provide account information or help someone understand a transaction.
• A fitness app might provide a workout or track an activity.
• A retail app might help someone find a product and place an order.

The interface is one way of delivering those capabilities, but it isn't the capability itself. That distinction becomes increasingly important when designing products for an AI-led ecosystem. Instead of asking only: “What screens should our app have?” Product teams may increasingly need to ask: “What can our product do, and how could someone ask for those things?” That's a much more interesting design question.

This isn't just theoretical product strategy. Apple's developer guidance gives us a pretty clear indication of the direction of travel.

With App Intents, developers can describe the actions their apps perform and the entities those actions operate on. Apple Intelligence can then use that information to understand the app's capabilities. Apple specifically recommends identifying the actions people perform most often and considering the contexts in which those actions are useful. There's another subtle but important piece here: language.

Apple's documentation recommends using familiar terms for app content and actions because those terms help people interact with the product naturally through Siri. In other words, the way you describe your product's capabilities matters. That's something we already think about when we're designing digital products.

A feature might have one name internally, another name in the database and something completely different in the interface because that's the language users naturally understand. When AI is involved, that becomes even more important.

If your system calls something a “journey object” but your customer calls it a “booking”, you've got a language problem to solve. The machine needs to understand the structure of the product, while the person needs to be able to describe what they want naturally. Good product architecture sits in the middle.

Apple isn't alone. Google is also developing ways for Android apps to expose their capabilities to AI agents and assistants.

Android's AppFunctions API is currently described by Google as an experimental preview that allows apps to provide functions, data and actions to the Android operating system so they can be used by proactive features, agents and assistants such as Gemini. Google describes AppFunctions as a mobile equivalent of tools within the Model Context Protocol, or MCP.

There is an important caveat here: Google's documentation says that AppFunctions integration with Gemini was in private preview as of May 2026. So this shouldn't be interpreted as saying that every Android app can already be freely controlled by Gemini in the way described above. What it does show is where the platform itself is heading. The operating system is starting to provide mechanisms for apps to expose functions rather than simply presenting an interface.

That distinction matters. A traditional integration might say:“Here's a button that opens our app.” An AI-oriented integration starts to look more like: “Here's an action our product can perform.” That is a much more flexible building block.

There is another reason this matters, App stores are relatively static environments. You search for something, and you receive results. AI assistants can potentially understand far more context around the request.

Consider these two searches: “Restaurant booking app” and: “Find me somewhere for dinner near the station tonight. I'm meeting two colleagues, one is vegetarian, and I'd rather somewhere we can actually talk.”

The second request contains considerably more information. An assistant can potentially use the person's location, time, preferences, previous behaviour, calendar and other context to determine what they actually need.

Apple's current Apple Intelligence documentation specifically talks about personal context, semantic search, app entities and contextual awareness. Developers can make app content available to the system and provide information about what is happening within an app so that Siri can better understand references such as “this photo” or “that message”.

This moves discovery away from a simple list of keywords and towards something much closer to contextual matching. And that changes what it means to be discoverable.

We're not suggesting that the App Store suddenly becomes irrelevant. There will still be plenty of reasons to browse an app marketplace. People will continue discovering brands, comparing products, reading reviews and deliberately looking for apps.

But there could be another route into the product… the conversation. Someone might ask an assistant: “What's the best way to keep track of my medication?” The assistant might explain a few options, they might then say: “Can you help me set that up?”

At that point, the interaction could potentially move from recommendation to action. The user hasn't necessarily searched for your app, they haven't necessarily seen your App Store screenshots and they haven't necessarily read your carefully crafted app description. Instead, the assistant has effectively become the discovery layer. That creates an entirely different competitive environment.

It's worth being clear about this because it would be easy to overreact. Your App Store listing still matters. Your Google Play listing still matters. Your website still matters. Your SEO still matters. Reviews, ratings, brand recognition, performance, accessibility and a good onboarding experience still matter too.

In fact, some of these things may become even more important because AI systems need reliable information from somewhere. What changes is that they are no longer the only discovery surfaces you need to think about. We've already seen a similar transition on the web.

Companies used to think primarily about ranking on traditional search engines. Then search became more conversational, featured snippets became more important, and now generative AI systems can summarise information and recommend businesses directly.

Adobe's research has found rapid growth in AI-driven referral traffic and argues that appearing in generative AI responses is increasingly becoming part of the consumer consideration journey. The same principle can start to apply to apps. You need to think about how your product can be understood, recommended and accessed by systems sitting between your customer and your interface.

This is where the conversation becomes much more practical. If you're currently planning an app, you don't need to throw your existing product strategy in the bin and start building an AI agent tomorrow. What you do need is to think beyond the interface. Start with the jobs your product actually helps people complete, not the screens, not the features, not the technology. The jobs.

If your app allows someone to book a service, what exactly does “book a service” involve?

• What information does the user need to provide?
• What decisions do they make?
• Which parts can be automated?
• Which actions need confirmation?
• What information does the system need to retrieve?
• What happens when there are multiple possible results?
• What happens when the user's request is ambiguous?
• And which parts of that journey genuinely require the person to open the app?

These questions are useful even if AI never enters the picture, because they force you to understand what your product actually does. But they also create the foundations for agentic experiences later.

One of the biggest mistakes we see in product development is starting with screens because they're tangible. It's much easier to draw a login screen than it is to define the underlying product logic. But when you start thinking about AI interactions, the limitations of a screen-first approach become obvious.

An assistant doesn't necessarily need to know that your booking button is blue and sits in the top-right corner. It needs to understand:

• what a booking is
• how to find one
• how to create one
• what information is required
• what options are available
• what actions are safe to perform automatically
• which actions require confirmation
• what happens when something goes wrong

That is product thinking rather than interface thinking. And it's something we put a lot of emphasis on during discovery.

Before we start building, we want to understand the problem, the audience, the journeys and the requirements properly. That gives us a much stronger foundation for deciding which experiences belong in the interface, which should be automated and where AI could genuinely add value. The technology comes afterwards.

A useful exercise for any product team is to take your existing app and temporarily ignore the screens. Instead, make a list of the things a user can actually do. For example:

A travel product might allow someone to:

• Search for journeys
• Compare departure times
• Check availability
• Buy a ticket
• View a booking
• Change a journey
• Cancel a booking
• Download a ticket
• Get disruption information

A finance product might allow someone to:

• Check a balance
• Search transactions
• Categorise spending
• Move money
• Set a payment
• Freeze a card
• Find a statement
• Understand a transaction

A field service product might allow someone to:

• Create a job
• Assign an engineer
• Check availability
• View a route
• Update job status
• Add notes
• Upload evidence
• Notify a customer

Suddenly, the product looks very different. Instead of a collection of screens, you have a collection of capabilities. And those capabilities are much easier to reason about in an AI-powered world.

This is where good product thinking becomes particularly important. It would be tempting to make everything executable through an assistant. But convenience isn't the only consideration.

Imagine an AI assistant being able to move money, cancel a booking, delete an account or send a sensitive message without any confirmation. The technology might be capable of doing it, but that doesn't automatically make it a good experience.

Apple's current App Intents guidance explicitly includes mechanisms for confirmation around destructive or sensitive actions and ownership of entities.

That's an important principle for anyone building an AI-enabled product. The question isn't simply “can the AI do this?” It's: “Should the AI do this, and under what conditions?”

Some actions can be safely automated. Some should require confirmation. Some might need additional authentication. Some should remain entirely inside the product's existing interface. That needs to be considered during product design rather than bolted on after the AI has been introduced.

There's another interesting consequence. For years, digital products have accumulated features, such as:

• More dashboards.
• More menus.
• More settings.
• More tabs.
• More notifications.
• More things competing for attention.

But if people can simply ask for something directly, some of those journeys can become considerably shorter.

Instead of opening an app, finding the right section, selecting an item and completing a series of steps, the user might be able to express what they want in one natural request. That doesn't mean the interface disappears. There will always be situations where visual interaction is better. Complex decisions, browsing, comparison, exploration and reassurance are all areas where a screen can be incredibly useful.

But it does mean we can start asking whether every action deserves its own journey through the UI. Some things might be better handled conversationally, some might be better automated, some might work better through voice, some might happen through notifications, widgets or wearable devices, and some will still belong exactly where they are today: inside the app.

The future isn't necessarily about replacing one interface with another. It's about having the right interface for the job.

This creates a slightly uncomfortable question for product owners. If an AI assistant becomes the interface between your customer and your product, how much of your brand experience actually survives? A beautifully designed app can communicate a huge amount through visual hierarchy, animation, typography, tone and interaction.

A conversational assistant can't necessarily reproduce all of that. So your brand needs to become clearer at the level of the product itself.

• What does your service sound like?
• How does it respond when something goes wrong?
• What information does it prioritise?
• How does it explain decisions?
• When does it ask for confirmation?
• How does it handle uncertainty?
• What does it do when it doesn't know the answer?

These are becoming part of product design too. The experience of using your product might eventually begin before someone sees your interface and continue after they've left it.

Today, many teams measure things like:

Download → open → sign up → screen → screen → conversion.

That's useful, but an AI-led product ecosystem creates additional questions.

• How often is your capability being invoked?
• Which requests are people making?
• Which requests fail?
• Where does the assistant need clarification?
• Which actions are completed successfully?
• Which actions result in the user being sent into the app?
• Which tasks are now completed without opening the app at all?

That last metric could feel strange at first. After all, product teams have spent years trying to increase app engagement. But if someone can complete exactly what they need without opening six screens, that might actually be a better product experience. The goal isn't necessarily to maximise the amount of time someone spends inside your app. The goal is to help them accomplish something valuable.

That distinction is going to become increasingly important as AI removes friction from digital experiences.

It would be easy to read all of this and assume that we're heading towards a world where everyone simply talks to an AI and apps disappear into the background. We're not there. There are still significant questions around privacy, permissions, reliability, authentication, commercial relationships, platform access and user trust.

And the underlying technologies are developing quickly. Google's AppFunctions, for example, is still described as an experimental preview, while Apple's capabilities and developer frameworks are continuing to evolve. There will also be plenty of situations where people deliberately want to use an app.

If you're booking a complicated holiday, managing your investments, designing a room or browsing a huge catalogue, there is real value in having a rich visual environment where you can explore and make decisions. So we don't think the future is “apps versus AI”. It's more interesting than that, it's apps plus AI plus whatever interface makes the most sense at the time.

This is the shift we think product owners should pay attention to. An app used to be the product. Increasingly, an app is becoming one part of a product ecosystem. The same underlying service might be available through:

• A traditional mobile app
• A website
• Voice
• An AI assistant
• A wearable
• A widget
• Notifications
• Messaging
• An API
• An automated workflow
• An AI agent

The customer doesn't necessarily care which one they're using, they care that they can get something done. That means the job of a product team is becoming less about designing a single interface and more about designing a digital service that can work across different interfaces. And that is a very different way of approaching product development.

If you're already running an app, start by looking at the actions that matter most. Here are some questions to ask yourself:

• What do people repeatedly come into your product to do?
• Which of those actions are relatively simple?
• Which require lots of navigation?
• Which involve information that could be retrieved automatically?
• Which could benefit from natural language?
• Which would be useful outside the app?
• Then look at the structure underneath.
• Are those capabilities exposed cleanly through APIs?
• Are your data models well defined?
• Can individual actions be triggered independently of the UI?
• Do you have clear permissions around what can and cannot happen?
• Do your systems know the difference between searching, previewing and committing an action?

If you're planning a new product, this thinking can happen even earlier. During discovery, you can map the user's jobs, define the underlying capabilities and identify where traditional interfaces, automation and AI each make sense. That can save you from building an experience around assumptions that may look increasingly outdated as the platforms around it evolve.

This is perhaps the most exciting part of where all of this is heading. We have spent decades designing digital products around the idea that a person opens something, navigates it and performs an action. Now we're starting to build products where the person can simply describe what they want and the technology works out how to make it happen.

The app doesn't disappear. The value it provides doesn't disappear. The need for thoughtful design certainly doesn't disappear. What changes is where the interaction begins.

It might begin with Siri or Gemini, it might begin with a voice command, a notification, a wearable or an AI agent, and sometimes it will still begin with someone tapping your icon because they simply want to explore your product themselves.

For product owners, that means the question isn't just: “How do we get people to download our app?” It is becoming: “How do we make our product useful wherever our customers are asking for help?” That's a much bigger opportunity.

And, from our perspective, it's exactly why the best product development starts with understanding the problem rather than jumping straight into the interface. If you understand what people are trying to achieve, the technology can evolve around that goal, whether the final experience is an app, a website, an AI agent, a voice interaction or something we haven't quite imagined yet.

Because the next generation of apps may not always ask people to find the app first, they may simply help them get what they need.

Ready to think beyond the app? The way people discover and interact with digital products is changing. If you're exploring what AI, automation or new interfaces could mean for your product, our team can help you understand where the opportunities are and what it would take to turn them into something real.