The advent of AI, and specifically agents, has shifted the “home screen” of consumer software from icons and grids to chat bubbles and threads. We are quickly moving away from thinking of “the app” as something you download from the App Store and tap through. The surface area of the internet is shifting, and the application layer is no longer a destination; it’s a conversation. Consumers don’t have the time or desire to “point-and-click"; rather, they want the intuitive and frictionless experience of  "describe-and-done."

The core shift is simple: instead of endlessly tapping through screens to find that one health data point you want from your Whoop or dig through emails to unsubscribe from that pesky subscription you keep forgetting about, you describe what you want and an AI agent orchestrates the work across your apps, accounts, and data to retrieve the information and complete the action with minimal effort. The core experience of that application’s interaction with the user is no longer the native app UI; it’s whatever chat rail the user already lives in: SMS, iMessage, WhatsApp, Telegram, Discord.​​

Think about your favorite budgeting app: instead of forcing you to open a dashboard, it could text you, “You’re on pace to overspend at restaurants by 32% this month; want to move $200 from your savings account to your checking and cut one dinner out?” and handle the money movement if you reply “yes.” Your favorite health app could ping you in the afternoon: “You’ve only walked 3,000 steps; want a 12‑minute walk and mobility session?” and adjust your plan when you text back “make it 8 minutes, I have a call.”

In this world, the “app” is a companion to the chat experience, not the other way around.

If you want to see the end state, look at WeChat. WeChat mini programs are essentially apps within the messaging environment, covering communication, shopping, bill payments, news, and bookings, serving more than a billion monthly users. Ordering food, paying utilities, booking doctor appointments, hailing a ride—much of it happens inside a chat super-app, where “installing” a new service is closer to opening a link than going to an app store.

That model was always culturally and infrastructurally ahead of the West, but AI agents are what finally dragged US users in the same direction. Once you realize you can just text an assistant to “book me a cleaning service this Friday, 3–5 pm, $120 budget,” instead of juggling three apps and a dozen taps, your expectations reset permanently. There is very little consumer patience left for front ends that force you to relearn navigation patterns with each new service.

AI didn’t invent messaging-native UX; it made it inevitable in markets previously attached to app stores.

OpenClaw has been a key catalyst for this shift and is a core reason it is happening now rather than five years ago. Never before could an agent running on messaging rails understand your intent with such accuracy and execute nontrivial, complex workflows on your behalf. OpenClaw displayed the power AI agents can have across messaging rails when provided real system access (terminal commands, browser automation, file operations, GitHub monitoring). 

People are experimenting with using OpenClaw-style setups to:

The important change is that chat stops being just a text interface and becomes a remote control for a general-purpose operating system. As that continues to come to fruition, building a rich application-based front end for everything starts to feel like unnecessary friction.

Many interesting new products are explicitly “messaging first,” with native apps playing a supporting role, if any role at all. Early messaging-native products:

A few concrete examples:

The big unlock for messaging experiences has been infrastructure that lets agents behave like first-class citizens on messaging rails.

Linq, for example, provides communications APIs for iMessage, RCS, SMS, and voice, including emoji reactions, voice notes, rich media, typing indicators, and group chats. They power many of the messaging-native assistants listed above, as well as companies that have ditched their mobile apps entirely, moving all customer workflows into chat. 

Instead of just sending/receiving messages, bots can now:

The result is that messaging-based experiences feel more humanized. Your interface to applications stops being a static grid of icons and becomes an ambient presence in your existing social channels, capable of the same modalities your friends use.

A Poke built on Linq doesn’t just answer questions; it can quietly join your family text thread to coordinate travel, jump into a group chat with your gym buddies to schedule a session, or follow up a complex conversation with a concise voice note recap. That is a very different feel from tapping around a cold SaaS dashboard.

RCS (Rich Communication Services) supercharges SMS like experiences with many of the capabilities we used to associate with standalone apps, such as richer media, improved group messaging, read receipts, typing indicators, and interactive elements. 

RCS makes it trivial for an AI agent to:

Combine RCS with AI, and you don’t just get a better SMS experience, but a full-featured application runtime that happens to speak in text, photos, files, locations, and voice notes.

Imagine texting your agent: “We landed in Austin; find a great taco place within a 10‑minute walk, kid-friendly, no reservations needed.” You share your live location in RCS/iMessage, it pulls up candidates, sends a tappable card for each, and when you reply with a thumbs-up emoji, it starts navigation and pings the group chat with the destination.

You never opened a travel, reservation, or maps app directly; you just chatted.

Most product teams still frame chat as text back and forth with an AI. That’s no longer true. Modern SMS, iMessage, WhatsApp, and RCS have practically all the sensory inputs of a native app:

One of the most powerful UX patterns I’ve seen is location and intent in one message: “Here’s my pin, find a quiet coffee shop within 7 minutes that won’t mind if I take two Zoom calls.” The agent can combine maps, reviews, noise-level data, and your past preferences to recommend options, then drop a navigation link back into the thread.

The chat thread becomes a command line for your life, but with photos, audio, GPS, and social context included by default.

Where attention goes, advertising follows. As more usage shifts into chat-based AI experiences, we should fully expect ads to move into those contexts.

Companies like Koah are already building GenAI ad networks that embed contextual ads directly inside AI conversations, helping chat apps monetize without jamming banners or interruptive interstitials into the flow. Koah’s pitch is essentially “AdSense for chat”: context-aware, native-feeling placements that appear when relevant—e.g., you’re planning a trip and a hotel offer appears inline that you can book with a single tap.

Kontext is experimenting with ad rails for apps that embed chat. Think assistants inside banking or shopping apps where conversation drives product discovery, and ads are woven into that dialog rather than bolted onto a sidebar. 

It’s not hard to see the next step- these same ad networks following AI agents into iMessage, WhatsApp, Telegram, and RCS, where your conversational intent becomes the most valuable signal for high-intent, performance-driven offers.

The underlying trajectory of ads that are conversational, contextual, and triggered by explicit user intent feels inevitable.

At a minimum, native apps’ role shifts dramatically, and they could very well disappear entirely over the long run. In the short run, I see it playing out as follows:

The home screen has been evicted by the AI overlord (I mean landlord 🙂), and the most important consumer digital real estate space is quickly becoming the top of a consumer’s favorite messaging app.