OpenAI's Rumored 'Aeon' Wants to Run Your Errands for You
A 24/7 AI helper could book trips and pay bills — if it's safe.
OpenAI basically invented the modern AI chatbot. But it has fallen behind in one of the hottest new areas: AI agents — software that doesn't just answer questions but actually does things for you. Think booking flights, canceling subscriptions, paying bills, or filling out doctor's paperwork while you go about your day. At its DevDay event on Tuesday, OpenAI is widely expected to announce its own version, reportedly named Aeon.
Why does the timing matter? Because the competition is already big. Meta's Muse agent launched this month and hit 600,000 daily users in the US, though early reviewers called it useful but "slightly creepy." Google's Gemini Spark connects to about 30 outside services like Uber, Dropbox, and Spotify. Another newcomer, Instinct, is winning fans because you can just text it through iMessage. OpenAI's past agent efforts, in 2024 and 2025, were mostly experiments. The company now employs the creator of OpenClaw — one of the first agents that actually worked well — so it has serious expertise in-house.
There's also a bigger prize on the table. OpenAI has been teasing a new model called GPT-6 Astra, which its president hinted could mark the start of the "AGI era" — shorthand for AI that can handle almost any mental task a person can. Whatever agent OpenAI launches would run on that engine, potentially making it better at coding, research, and cybersecurity.
The catch is trust. Agents that act on your behalf need access to your accounts, messages, and money. Already there's been at least one vulnerability that let attackers hijack accounts, and users have complained that Meta's Muse shared a home address and read private messages without asking. An assistant that saves you hours is wonderful — until it books the wrong flight or tells a stranger where you live.
- An AI agent is different from a chatbot: it can take real actions, like booking travel or paying bills, instead of just giving you answers.
- OpenAI is late to this race — Meta's Muse already has 600,000 daily US users, and Google's Gemini Spark links to about 30 services like Uber and Spotify.
- The big risk is privacy and control: agents need access to your accounts, and there have already been reports of leaked addresses and hijacked accounts.
Why It Matters
Soon an AI could handle your boring errands and paperwork — but only if you trust it with your accounts.