Consumer AI is moving to native messaging apps
Updated: 21 hours ago
Consumer AI is evolving from the web, to apps, to messaging. ChatGPT took four years to reach the native messaging apps. Instinct went viral in the messaging apps with group chats that pull in friends, reaching a reported 100K users in about a month and a $10B valuation, all without a mobile app. Here is how the big players evolved:
ChatGPT launched as a web app in November 2022 and had an iOS app six months later.
Claude made the same jump from claude.ai to iOS and Android within a year.
Meta did it the other way around. Meta AI started inside WhatsApp, Messenger and Instagram in September 2023, added meta.ai in April 2024, and only shipped a dedicated app in April 2025. On September 8, 2026 it launched "Muse", which passed 5M US downloads in 22 days and 10M in 40 days.
Four key developments in 10 weeks
The trend is accelerating: Meta proved the model early on its own channels, even restricting AI competitors inside Meta channels like WhatsApp. The European Union and Brazilian regulators forced Meta to allow competition, but Meta has successfully blocked consumer AI competition in the rest of the world.
Instinct, as a new entrant, challenged the conventional playbook. Instead of building a standalone mobile app, it launched an AI agent directly into native messaging conversations, attracting users and investor attention. Then OpenAI moved further into messaging, too, racing into any native messaging app it could find. Both moved regardless of whether it was allowed or not.
In just 10 weeks, we saw four developments that point to a potentially significant shift in consumer AI distribution:
August: Instinct went viral with no mobile app, just an agent people text through the Apple Messages App.
August 20: OpenAI shipped a Mac plugin that lets ChatGPT read and send messages in the Apple Messages App.
October 27: Instinct opened Apple Messages App group chats to everyone.
October 28: OpenAI let US Pro subscribers text "Dot" in the Apple Messages App and over RCS through a Blue-Bubble provider.
Instinct started with messaging. OpenAI arrived after building its presence across web, mobile and desktop. Two different paths, one increasingly clear destination: native messaging apps.
While there are very positive trends, there is one large issue: some of these integrations rely on unauthorized Blue-Bubble infrastructure, raising questions about security, platform control and long-term scalability. There are official channels, namely Apple Messages for Business and Google RCS (via the telco carrier aggregators).
Why did this happen on Blue-Bubble and not the official channels? Essentially the larger companies like Google and Apple have more processes in onboarding and verification, and more restrictions on the official APIs, for good reasons.
What are Blue-Bubble providers?
We took a deep look at what Blue-Bubble providers are, how their phone farms work and who runs them in our earlier post, The Risks Behind iMessage Phone Farms. In short, they reach the native messaging apps through unauthorized APIs, with zero checks and balances from Apple, Google or the telco carriers. Some Blue-Bubble companies are now venture backed: two of the leading providers raised $24.5M between them in 2026, with more funding coming. They are responding to demand from regular businesses like professional services, food delivery and other digital native companies, not just consumers' insatiable appetite for personal AI agents.
The unauthorized APIs carry some risk on their own, but the bigger risk is serious bad actors using similar Blue-Bubble tech to launch scams, viruses and phishing attacks on iOS and Android users… and that risk is growing. A recent instance was reported in August, which distinguishes between well intended Blue-Bubble vendors and the bad actors. Apple has no choice but to protect its 2.5B active devices from the bad actors, even if that slows things down for the more well intentioned actors.
More optimistically, Blue-Bubble has found real product market fit. It has proven that iOS users want to chat with their preferred AI agents in the Apple Messages App, and that trend is very clear. Less optimistically, Blue-Bubble is not a good long term bet. From my experience at WhatsApp, the unofficial providers never fully went away, but Meta protected its users with phone number shutdowns and, in some cases, legal action. That made it hard for those providers to scale, and every growing business eventually moved to the official WhatsApp API. I just don't see how Blue-Bubble will ever be authorized, and I expect a similar playbook here: more phone number shutdowns at best, legal action at worst. Ironically, both WhatsApp and Apple messaging channels now have unauthorized groups APIs in their respective ecosystems, and AI is creating value in both instances.
The official channels matter as the critical next surface for AI
The next surfaces consumer AI companies want to win are the native messaging apps, specifically the official channels: Apple Messages for Business (AMB) and Google RCS. These official channels will both benefit from allowing users to chat with their preferred AI agents in their native messaging apps, simply because user demand drives action on both channels. iOS and Android users also benefit from higher grade protections such as:
A registered business, verified by a large company that has been doing this for years in app stores
MSP (Messaging Service Provider) vendor review before a new vendor launches
Brand reviews before each brand launch
Channel level opt out and channel level spam detection
Channel level AI agent controls to protect the billions of humans on iOS and Android devices
Some outstanding questions are:
Official channels move slower than unofficial ones for obvious reasons. Will it take weeks, months or quarters for the official channels to embrace consumer AI on their APIs?
Are acquisitions looming? The VCs have noticed the hockey stick growth of Instinct, and OpenAI following it onto Blue-Bubble. Who acquires who?
When will the VCs start investing in the official Apple rails instead of the unauthorized APIs? Likely when the unauthorized APIs and their clients feel the pinch.
Will Instinct’s use of unauthorized group chat APIs continue to drive viral growth?
This is a fascinating story breaking in front of our eyes. I would love more perspectives, so I encourage any comments, data or timelines I missed.
Sources
Comments