- IoT For All
- Posts
- OpenAI's ChatGPT Agent wants to do it for you
OpenAI's ChatGPT Agent wants to do it for you
And our top AI story of the week
Hello readers,
Welcome to the AI For All newsletter! Today, we’ll be going over OpenAI’s new “ChatGPT Agent” feature, how AI is helping make manufacturing more efficient, and more!
AI Spotlight: ChatGPT Agent

What: Announced yesterday, ChatGPT Agent is OpenAI’s biggest leap yet toward turning its flagship chatbot into a full-blown digital assistant. This new “agentic” mode gives ChatGPT its own virtual computer complete with a browser, terminal, and access to apps like Gmail and Google Calendar so it can carry out complex tasks for you, end to end.
How: You activate Agent Mode from within any ChatGPT conversation (available to Pro, Plus, and Team users), and describe the task you want it to do. From there, it’ll pick the right tools and start working. It asks for permission before taking any consequential actions, and you can pause, jump in, or take over at any point. Tasks persist across multiple steps and tools, meaning ChatGPT can gather data from a webpage, run code to process it, then present the output in a downloadable format, like a spreadsheet or presentation. You’ll see everything it’s doing through live narration and progress updates.
Why: This is OpenAI’s answer to the “do-it-for-me” era of AI. Instead of being limited to conversation, ChatGPT Agent can now think and act, making it vastly more useful for everything from professional research to life admin. It’s a natural evolution of OpenAI’s Operator and deep research tools, combining real-time web interaction with advanced reasoning. Early users can oversee tasks like preparing slide decks, editing spreadsheets, and summarizing inboxes without jumping between apps or writing a single line of code. Yes, there are still some rough edges (like basic slideshow formatting), but there’s a lot of potential.
Our favorite part: The combination of deep research and real-world action. It’s one thing to ask an LLM to explain something, it’s another to ask it to do the work, then have it actually navigate the web, gather files, and return editable outputs like a real assistant would.
Pricing:
Pro: $20/month includes 400 messages/month with full Agent access
Plus and Team: 40 messages/month with Agent access included
Enterprise and Education: Rolling out soon
Deals or promotions:
No special promos, but Pro users get the earliest access and highest usage caps.
Disclaimer: LLM-based agents often don’t function as intended.
🔥 Rapid Fire
Why some companies are canceling ChatGPT subscriptions
Musk's Grok signs $200M deal with Pentagon days after antisemitism row
Is AI the new Android bloatware? Why phone-makers are scrambling for your attention
Will AI end cheap flights? Critics attack Delta’s “predatory” AI pricing
Meta appoints generative AI VP to run Threads
No, AI is not the reason for mass layoffs
AI firms ‘unprepared’ for dangers of building human-level systems, report warns
Where are all the AI drugs?
How Nvidia’s Jensen Huang persuaded Trump to sell A.I. chips to China
Lavish spending and weak growth engulf AI startup SandboxAQ
Perplexity’s CEO claims the browser is AI’s killer app
📖 What We’re Reading
“Operators are juggling multiple jobs. Engineers are drowning in dashboards. And managers? They’re stuck trying to interpret a sea of KPIs with delivery dates breathing down their necks. What used to be a staffing problem is now something deeper: how do we act on the flood of information we're already collecting without burning out the few skilled people we have?“