OpenAI introduces ChatGPT agent

💥 Announcing ChatGPT agent: a powerful new agent that can use a computer, browse the web, write code, use a terminal, write reports, create images, edit spreadsheets, and even create slides for you. The slides often... need some work. But you know how this goes: first it's impossible, then it just kinda works, and then very quickly it's great and we never look back. ChatGPT agent is beginning to roll out today to Pro, Plus, and Teams users, coming soon to Enterprise/Edu.

OpenAI’s announcement about ChatGPT Agent is really exciting. To me, it represents a big step forward in how we’ll all begin to use AI to accomplish day-to-day tasks. So as someone who wants to save time and use AI to do that, I’m thrilled. As someone who is passionate about cybersecurity, I do have a few questions: 1. This agent would have access to ALL the data on your computer. How do users know what the company is doing with that information? 2. How will ChatGPT Agent navigate the often-blurred line between work and personal emails, calendars and documents? 3. Most importantly, can users audit what the desktop agent is actually doing, and accessing? It’s clear how ChatGPT Agent benefits OpenAI; it’s a desktop app which could give the company access to massive amounts of your data. This is akin to the early advantage Google received when it won the search wars and built a marketing machine with all of our search data. That puts OpenAI in a powerful position. It will be even more powerful if customers have full trust that their data and personal information is secure, and being used for exactly the purposes they intend. Full stop. I believe innovation in AI and progress in security must go hand in hand. We can build tools that are both incredible and accountable — and I’m working on something now that will help make that possible. For companies innovating in AI: it’s never too early to bring in a cybersecurity partner. More soon. See a full article about ChatGPT Agent at the link in my comments.

Here's a peek into the future. ChatGPT just launched an agent that has full access to its own computer. This means that it can access far more tools and run programs in the background to do work. When combined with your existing corporate information, this becomes insanely powerful for doing knowledge work. Here's an example of what that looks like with the ChatGPT agent connected to Box as a data source with the task of creating a powerpoint presentation using market research content within Box. The agent has to go find the most relevant information for the presentation, read through the documents to find the important insights, and use a computer to create a powerpoint presentation. This would have been impossible just a year ago. The big jump in AI over the past year has been moving from AI that you chat back and forth with to get work done, to a paradigm of AI Agents that you can send tasks off to and they go and complete work for you. Agents can go off for hours and complete tasks for us, and will just notify us when they're done doing their work. This is a small example of what that will look like in the future.

Would you ditch your browser for an agent? OpenAI is hoping you will. OpenAI has launched ChatGPT agent, which can control your whole computer and perform multistep tasks, powered by a new dedicated model: https://lnkd.in/gPXGmnwm ChatGPT agent is supposed to be able to do anything you can do on a computer on your behalf, but mainly it just stumbles around. That doesn’t mean that it’s useless: you just need to find a task that it does well and is useful to you. ChatGPT agent is available to ChatGPT Pro, Plus, and Team subscribers by selecting “agent mode” at any point in any chat. While Google and Microsoft are adding AI features to their browsers and Perplexity and The Browser Company are launching AI-first browsers, OpenAI is abstracting the browser away entirely. Will it work? It’s way too early to tell: agents will likely make countless mistakes, but as long as AI companies keep investing in them, users will find some use cases.

No better time to dig into this topic, with OpenAI releasing ChatGPT Agent Mode today, and Perplexity rolling out its agentic browser Comet last week. Here's why there’s a sudden flurry of investment and experimentation in browsers, what is at stake and who stands to win and lose as these challengers take on incumbents like Google. We're officially entering the latest iteration of the browser wars! 🔥

Happy OpenAI Launch Day to those who celebrate! This morning, OpenAI hosted a good ol’ fashioned livestream and launched ChatGPT Agent, marking its biggest shift yet from assistant that answers to agent that acts. The pitch is simple: describe a task in plain English, and it does the thing. Not just answering questions - actually doing. The agent clicks through websites, fills out forms, runs code, checks your calendar, generates editable PowerPoints and spreadsheets - and circles back only when it needs approval. The agent now handles the tedious 80%. It fuses OpenAI’s past work - Operator (browser automation) and Deep Research (synthesis engine) - into one persistent, goal-completing system. Think of it as ChatGPT with a mouse, memory, and a working knowledge of Excel. OpenAI says it’s especially good at “first-year analyst” work: compiling competitive landscapes, summarizing earnings calls, maybe even drafting that 20-slide board update you were going to pull an all-nighter on. If this feels like a shot across the bow at Microsoft… well, it is. But OpenAI insists it’s not. “This is just how people communicate,” said OpenAI’s Chief Product Officer. “They create spreadsheets, they create PowerPoints.” Sure. And now, so does ChatGPT. So what’s actually new? ▪️It runs on a virtual computer. So it can interact with real software UIs, not just APIs ▪️It outputs fully editable files. Presentations and spreadsheets you can open in Microsoft Office (or Notion, if you’re feeling chaotic) ▪️It handles research-to-output workflows. Ask it to look something up → analyze it → summarize it → generate a slide deck ▪️It integrates with your apps. Gmail, GitHub, calendar, etc. (no financial actions… yet) Latency is still an issue (some tasks take 15–30 minutes), but OpenAI’s not optimizing for speed. It’s optimizing for replacement labor. With this launch, distribution shifts from API to behavior. This isn’t something a dev team integrates. It’s something you use directly. A personal software layer that lives inside your browser, listens to your prompts, and quietly takes work off your plate. Let's talk about safety. There are safeguards. Watch modes. Refusal training. Red teams. It prompts for permission before doing irreversible things, like bookings or purchases - for now. Still, we’re now in a world where you can say: "Plan my birthday, compare cake prices, and email me the top three vendors." - and it’ll do all of that. Autonomously. While you write a LinkedIn post about it. The dream of AI agents has been around forever. But today’s launch is productized, permissioned, and rolling out to Plus, Pro, and Team users immediately. OpenAI has officially graduated from offering inference as a service to execution as a product. If GPT was the brain, this is the body - clicking, typing, formatting. Like every OpenAI launch, the product will now get battle-tested in public now. Let the testing begin!

📰 ChatGPT Agent Is Here—and It Can Actually Get Things Done For You ChatGPT Agent can perform full workflows across the web: it clicks buttons, fills out forms, scrapes data, runs code, and even builds slide decks. You talk to it in natural language, it figures out what to do, and gets to work using a virtual computer that mimics how a person uses a browser and desktop. It integrates with tools like Gmail, Google Calendar, and APIs via connectors and is available now to Pro, Plus, and Team users. Please use this technology responsibly. There are MAJOR safety and security implications for a tool like this. OpenAI has done a LOT to try to add guardrails to it, but it introduces a major new attack surface that OpenAI is aware that they may never be able to fully anticipate and cover. That said, general purpose AI agents like this are incredible to see in action and will completely change how you work. https://lnkd.in/eSdX6nCs

I canceled my Manus subscription a while ago because it could never keep up on accuracy. Pretty skeptical about this ChatGPT Agent release, but according to OpenAI I'll have it by end of day today... so let's see what it can do. What they claim it can do: - Browse websites, click/scroll/type like a human - Create editable slideshows & spreadsheets from scratch - Run code, analyze data, conduct research - Handle full workflows: "plan Japanese breakfast for 4" & then shops for ingredients - Calendar integration: brief you on upcoming meetings based on recent news - Competitor analysis & auto-generate slide decks - Convert screenshots into presentations - Plan & book entire travel itineraries - Schedule appointments, rearrange meetings - Update financial models with new data Basically all the stuff I tried to get Manus to do for me in the past. We'll see if it lives up to the hype or if it's another case of impressive demos that fall apart in real use. https://lnkd.in/gPe46-Du

OpenAI Launches ChatGPT Agent with “Agent Mode”, combining browsing, deep research, and a virtual computer in one tool. Effectively merging the capabilities of its earlier tools (Operator + Deep Research) into one multitool. Aimed at streamlining everything from financial analysis to client briefings and deeper research reports. Sends a strong signal that AI agents are shifting from chat to real-world tasks. It also puts OpenAI directly in competition with Microsoft, Google and others in the intelligent productivity space. If you’re a Pro, Plus, or Team ChatGPT user, you now have an intelligent assistant that can proactively execute real tasks, within guardrails, saving you time and mental effort. Worth testing for streamlining workflows across research, planning, and routine tasks. I am constantly reminded that the world isn't changing, it changed. Now that change is growing faster than ever. #AI #OpenAI #ChatGPT #FutureOfWork #GenerativeAI #AgenticAI #EnterpriseAI #ChatGPTAgent #TechNews #AItools #TaskAutomation