🧠 OpenAI’s Big Agentic Move: From Assistants to Real Actors

In partnership with

The free newsletter making HR less lonely

The best HR advice comes from people who’ve been in the trenches.

That’s what this newsletter delivers.

I Hate it Here is your insider’s guide to surviving and thriving in HR, from someone who’s been there. It’s not about theory or buzzwords — it’s about practical, real-world advice for navigating everything from tricky managers to messy policies.

Every newsletter is written by Hebba Youssef — a Chief People Officer who’s seen it all and is here to share what actually works (and what doesn’t). We’re talking real talk, real strategies, and real support — all with a side of humor to keep you sane.

Because HR shouldn’t feel like a thankless job. And you shouldn’t feel alone in it.

Sign Up Free

There’s a quiet revolution happening inside OpenAI — and it’s not just about smarter chatbots. It’s about agentic AI — systems that don’t just answer but act. The company has been quietly laying the groundwork for AI that can browse, buy, build, and even patch software on its own. And in 2025, that groundwork finally looks ready.

Let’s break down what’s new, why it matters, and what it could mean for the rest of us.

⚙️ AgentKit — The Heart of the New OpenAI Era

Meet AgentKit, OpenAI’s latest and arguably most ambitious toolkit yet.
Think of it as an operating system for building AI agents — a one-stop suite that helps developers design, deploy, and manage agents without wrestling a dozen disconnected tools.

At the core of AgentKit are four big pieces:

• Agent Builder — a drag-and-drop visual canvas for creating multi-agent workflows. Developers can design complex behaviors, test them, and preview results without writing endless prompt chains.

• Connector Registry — an easy way to plug agents into real-world apps and data sources like Google Drive, Dropbox, or Slack — no extra APIs or code spaghetti.

• ChatKit — the bridge that turns these agents into chat-based UIs, so they can live naturally inside products, apps, or even customer-facing chat interfaces.

• Evaluation Tools — built-in systems for automated prompt optimization, trace grading, dataset testing, and integration with third-party evaluation platforms.

The goal? To end the messy “stitch-it-yourself” era of AI workflows. VentureBeat called it OpenAI’s effort to make “agent development accessible and robust for the next generation of builders.”

In short: OpenAI doesn’t just want developers to use ChatGPT — it wants them to create their own agents with OpenAI DNA.

🛍️ Agentic Commerce — “Buy It in ChatGPT”

If AgentKit is the engine, agentic commerce is the test drive.
OpenAI has rolled out a bold experiment: letting ChatGPT not just recommend products, but complete the purchase for you.

Dubbed “Buy it in ChatGPT”, the feature is now live in parts of the U.S., beginning with Etsy sellers. Imagine asking ChatGPT to find a handmade mug — and within seconds, it’s bought, paid, and shipped, all from inside the chat.

Behind the scenes, this is powered by the Agentic Commerce Protocol, co-developed with Stripe. It’s open-source and designed to let AI agents, users, and businesses safely coordinate transactions while keeping merchants as the official sellers.

This marks a massive shift — from assistants that recommend to agents that act on behalf of users.
And yes, that raises big questions about trust, consent, and control. But one thing’s clear: commerce will never be the same.

🧩 Evolution Beyond “Operator”

Remember Operator — that early OpenAI experiment that could browse the web, fill out forms, and even place orders?
Well, Operator walked so AgentKit could run.

Operator was essentially a research prototype. Now, OpenAI has folded its lessons into this new generation of agentic systems — more capable, safer, and deeply integrated into its ecosystem. It’s the natural evolution: moving from one autonomous agent to a full platform of them.

⚖️ The Pushback and the Questions

Not everyone’s clapping.
Hollywood, for one, is already sounding alarms. The Creative Artists Agency (CAA) recently warned that OpenAI’s video model Sora could threaten creators’ rights and compensation. The core question: If an AI agent generates or transforms creative work, who owns it?

It’s not just about movies — it’s about accountability.
As these systems act with greater autonomy, the risk of misuse, over-automation, or unintended actions grows too. When your agent clicks “Buy Now” or publishes content, who’s responsible for what happens next?

🔍 What It All Means

Agentic AI is no longer a futuristic concept — it’s here, and it’s scaling fast.
With AgentKit, OpenAI is betting that the next wave of innovation will come from developers who let AI do more than talk — AI that works, executes, and delivers outcomes.

But as autonomy increases, so does responsibility. The new frontier isn’t just technical — it’s ethical, legal, and economic. We’re entering an age where agents don’t just assist humans; they represent them. And that’s a whole new game.

Final thought:
The line between “assistant” and “actor” has officially blurred.
OpenAI isn’t just building models anymore — it’s building a workforce of digital agents ready to operate in the real world. The challenge now is making sure they do it safely, transparently, and in service of people, not the other way around.