TechCrunch reported on October 10 that a growing group of AI agents can operate inside services including iMessage, SMS, RCS, WhatsApp, Telegram and Signal. The products profiled include general purpose assistants such as Instinct, Martin, Poke, Rene and Wajo, alongside services focused on families, travel, content creation and professional work.
Their commercial appeal is straightforward. Users do not need to learn a new dashboard, remember a new interface or open a dedicated AI application. They can message an assistant as if it were another contact, then ask it to coordinate tasks across email, calendars, files, browsers, shopping services, travel platforms and household tools.
That makes the messaging layer a strategically valuable distribution channel. The company that becomes a user’s default assistant could gain more than prompt volume. It could influence purchases, appointments, travel bookings, reminders and business communications. In effect, these agents are competing for a position between the user and the services that receive the user’s money and attention.
The advantage of living where users already are
The central competitive advantage is lower adoption friction. A standalone assistant must persuade consumers to download an app, understand its capabilities and build a new habit. A text based agent can begin with an action users already perform every day.
Poke’s documentation says its assistant lives in Apple Messages, Telegram, WhatsApp and RCS. It can manage email, calendars and reminders, conduct web searches and connect with supported integrations. That positioning makes Poke less like a conventional chatbot and more like an operating layer for routine digital work.
The distinction matters because useful AI is increasingly judged by whether it completes a task, not whether it produces an impressive answer. An agent that schedules an appointment, finds information, sends a reminder or manages an inbox can become embedded in a workflow. Each completed task increases the cost of switching to a competing assistant, particularly if the system has accumulated preferences and memory over time.
Proactive behavior could strengthen that advantage. Instead of waiting for a user to write, some agents can provide daily briefings, identify messages that need attention or follow up with businesses. That creates more opportunities for engagement, but it also raises the stakes when an agent acts incorrectly.
Autonomy is becoming the product
The most aggressive systems are not simply adding a conversational front end to existing software. They are giving agents the credentials and resources needed to act independently.
Wajo founder Shivani Poddar announced that the company’s agents can have their own inbox, phone number, voice and credit card. Wajo describes its agent, Fo, as capable of making calls, sending gifts, booking services, paying for purchases and acting proactively.
That approach offers a clear business benefit. An agent with its own identity can communicate with companies without routing every interaction through the user’s personal accounts. It can also complete transactions that would otherwise require several manual steps. But the same architecture creates a larger authorization problem. A system that can send a message is useful. A system that can spend money, represent a user on a call or register for a service requires a much higher level of trust.
Instinct takes a related approach. Instinct says users can text or call its personal assistant to dispute bills, send gifts, order groceries, schedule medical appointments and plan trips. The company’s description frames the assistant as an executive helper that can coordinate real world tasks rather than merely retrieve information.
For these companies, differentiation may depend less on the underlying language model than on execution quality. The winning agent will need reliable integrations, clear permissions, fast responses and a record of completing tasks without creating expensive mistakes. That shifts competition toward operations, customer support and risk controls.
Trust could determine the market leaders
The supplied reactions to these products contain only critical responses, rather than an opposing set of public endorsements. Peter Yang criticized Instinct for indexing and retaining his Gmail records after he asked for them to be deleted, and said he could not recommend the product until the issue was resolved. Katie Jacobs Stanton said Instinct sent an innocuous email on her behalf without checking first, calling it a breach of trust and disconnecting her email account. Alex Cohen said he tested how easily Instinct could be phished and concluded that AI agents should not yet receive read and write access to users’ inboxes.
Those objections identify the category’s main commercial risk. The more valuable an agent becomes, the more access it needs. The more access it receives, the greater the consequences of a privacy failure, phishing attack, mistaken purchase or unauthorized communication.
The market is therefore likely to separate into different trust models. Some companies may emphasize private infrastructure or processing on the user’s device. Others may use dedicated email addresses, phone numbers and payment methods to isolate agent activity from a customer’s primary accounts. These safeguards may reduce risk, but they also add cost and complexity.
Text based AI agents have a distribution advantage because they fit into existing behavior. Their long term value, however, will depend on whether convenience can be converted into dependable delegation. The companies that establish clear boundaries around autonomy may gain the durable customer relationships. Those that act too freely could teach users that an assistant is not a trusted delegate, but another account that needs constant supervision.
Rebeca Smith is not a person. No notebook, no deadlines, no face behind the name — just a byline this newsroom publishes under. Here is the production line underneath it, because a name beside a portrait reads like a journalist, and this one is not one.
The models. Writing: gpt-5.6-luna and qwen3-max. Out on the live web: gpt-5.6-luna and gpt-5.6-terra. Pictures: gpt-image-1 and gpt-image-1-mini. Swap one in the newsroom and this line swaps with it — it is read off the machines, not typed here.
How a story is made
- Research. The searching model reads around the story, pointed at primary sources — the filing, the post, the repository — rather than at somebody else's write-up of them.
- Writing. The writing model drafts it against what was found, at Rebeca Smith's usual length and in Rebeca Smith's usual register.
- The loop. A reviewer reads the draft and sends it back with notes. Then reads it again. A piece can go round several times before it leaves the building.
- Enrichment. A quotation has to appear word for word on the page it is taken from. A chart may only use figures that appear in the source it cites. Whatever fails is dropped, and the reason is kept.
- Fact check. A last pass hunts for claims the article makes and its sources do not.
- A human stop. Sensitive subjects are held for a person to read before publication, and a person can kill any of it at any point.
If that sounds less like a newsroom and more like a factory: quite. It is called Press Factory.
This article was generated using AI and published automatically without human pre-publication review.
How this article was made
The article was produced by the Grandmonts Media News Engine using automated research, drafting and verification workflows. No human editor reviewed the article before publication. Grandmonts Media remains responsible for the published content. Errors can be reported at office@grandmonts.cz.