Your ChatGPT may soon be getting a new kind of team member. With Elon Musk's Bot having been on the job for over a month and Mark Zuckerberg's Muse climbing to the top of the App Store charts, competition is heating up. In the race to create AI agents that can handle real work, OpenAI has suddenly found itself in a defensive position after just a few short weeks.

OpenAI is reportedly developing a feature that directly rivals Grok Bot, according to a report from The Information. Internally, the company is also discussing how to respond to Meta's personal assistant, Muse. The label of "chaser" applied to OpenAI in this report would have seemed unthinkable just eighteen months ago, especially after it launched Operator in January 2025, becoming the first to make consumer-grade AI browser automation a reality. ChatGPT has since brought generative AI to everyday users, surpassing 10 billion weekly active users this summer according to the same report.

Following the news, Tibor Blaho, a developer known for digging into ChatGPT's code, surfaced a single word: Aeon. In Greek, Aeon translates to "an age" or "endless time." If this is indeed the name OpenAI has chosen for its agent, it strongly suggests the company is preparing to launch an AI colleague that works around the clock. Blaho posted a screenshot of code extracted from the ChatGPT client to support his claim.

The code found in the web version of ChatGPT reveals a key detail: a message sent by a user is attributed to the user's account, but a message not sent by the user is attributed to an entity named Aeon. This indicates that within ChatGPT's system, Aeon is treated as a distinct identity alongside the user. Even while being labeled a chaser, the leaked code shows a reserved "workspace" for this AI colleague within the app.

OpenAI's sense of urgency stems from the fact that Grok Bot has been operational for over a month. On August 11th, SpaceXAI launched Grok Bot with a simple mission: to be your always-on autonomous agent team. Rather than a single assistant, users are given a team of agents who possess their own dedicated computers, a key difference from standard AI assistants. These bots share a persistent cloud-based computer where files, browsers, and login states are stored, allowing them to continue working even when you close your laptop.

These bots can autonomously log into your Gmail, Notion, financial software, and even older websites without APIs. They operate like humans, using a mouse, filling out forms, and completing tasks end-to-end, only pausing to ask you for your approval on critical decisions. More impressive is their ability to collaborate. You can deploy multiple bots at once, assigning one to manage your inbox, another for outreach, and a chief-of-staff bot to oversee them all, with tasks being allocated among them in a group chat. The fundamental value of Grok Bot is in outsourcing the work.

The code screenshot shows an aeonId listed alongside accountUserId, suggesting Aeon could be an independently identifiable object within ChatGPT's framework, rather than a simple feature toggle, although the extent of its identity or permissions is not yet clear. The aeonId appears repeatedly in two recent consecutive Android builds, accompanied by a field named memberAeonIds. The term "member" hints at a format similar to Grok Bot's multi-agent group chat. Additional test documentation in Blaho's screenshot describes a scenario designed to protect "aeon-style behavior," requiring the system to recapture capability status at every sampling step rather than waiting for the next user prompt.

This is in stark contrast to standard ChatGPT's back-and-forth interaction, as "aeon-style behavior" requires the system to reassess its available capabilities after each step of a multi-step process. This logic matches the requirements of an agent that must execute a long sequence of actions. OpenAI has yet to make an official statement, leaving only a media report, a suspected codename, and a few code fragments for now. However, the pieces are starting to come together.

The most telling part is the report's claim that OpenAI will not build a new product from scratch but will instead "repurpose" the agent technologies already present in Codex and ChatGPT. This statement highlights an awkward truth: the capabilities competitors are actively marketing are already in OpenAI's possession. On July 9th, OpenAI released ChatGPT Work, which officially includes features like cross-app task execution, continuous processing of complex projects, and scheduled tasks, all while integrating Codex's technology. The official documentation also confirms that when using the cloud runtime, tasks continue to run even after you close the app or shut down your laptop.

Furthermore, the talent has been in place since February, when Peter Steinberger, creator of OpenClaw, joined OpenAI, with Sam Altman stating he would be responsible for building the next generation of personal agents. The problem is that many OpenAI users are unaware these capabilities exist. It is akin to a kitchen that can prepare any dish, but the menu only lists "chat." If the Aeon rumors are true, the real question is whether it will add a "digital member" that can be directly named and assigned tasks, building on the existing capabilities of ChatGPT Work. This would be OpenAI's direct answer to Grok Bot.

According to The Information, ChatGPT's growth has notably decelerated over the past year. OpenAI originally aimed to reach its 10 billion weekly active users target last year, but only hit it this summer. In this current wave, the competition is no longer about model leaderboard positions. AI agents that can complete tasks for users have become the second growth curve every major company is pursuing. Grok Bot is focused on a cloud computer holding all your login states. Anthropic was earlier with Claude Tag, launched on June 23rd, which integrates Claude as a member of Slack channels to execute tasks, targeting entire team workspaces. Meta's Muse reached the number 2 spot on the US App Store free chart within two days of its launch, and by September 22nd, it had hit number 1, pushing ChatGPT down in the rankings, as it aims to be your daily-used chat app.

Then there is Instinct, an invite-only startup whose valuation quadrupled within a month. It uses text messages and calls to book restaurants and make purchases, aiming to secure your credit card and passwords. These four competitors, with their distinct strategies, are all targeting your workflow. Picture a future Monday morning: just after 8 AM, you reach for your first coffee and glance at your phone. Previously, you would need to reply to a dozen emails, mute annoying group chats, and schedule a series of meetings. In the future, after finishing your coffee, these tasks might already be done, leaving you with a simple checklist to approve: send this email, pay this bill, accept this meeting.

Whoever controls that checklist holds the gateway to your daily work. Once this gateway becomes a habit, it's hard to switch. If a bot manages your inbox, learns your email tone, and knows which clients to prioritize, replacing it becomes far more difficult than changing a chat app. However, the biggest hurdle may lie outside technology: trust. How many everyday users are willing to grant an AI access to their personal information, social accounts, and even authorize it to make payments? This hesitation is the final barrier every "AI colleague" must overcome. In the chat era, users asked if AI was intelligent. In the agent era, they ask if they dare trust it with their work. The only question left for OpenAI is how to convince 10 billion users to hand over the keys.