This week, three stories show the same trend: AI agents are moving from "experimental tech" to "normal work tool." OpenAI says its users are running agents like full-time teammates. Samsung just gave ChatGPT to every employee. And in Korea, ChatGPT is coming to everyday group chats.
The shift is not about models getting smarter this week. It is about more people actually using them.
1. OpenAI Says Agents Are Becoming Normal Work Tools
What happened: OpenAI published a report on June 25 showing how people are using its Codex agent tool. Codex now has over 3 million weekly users. The most active users are running more than 60 hours of agent sessions per day, spread across multiple parallel agents.
Why it matters: 60 hours of agent time per day means these users are not just testing the tool. They are running it like a team member that works alongside them continuously. And Codex is expanding beyond software developers — it is being used for routine knowledge work across many professions.
Ask your Ally: "What routine tasks do I do every week that could be handled by an AI agent working alongside me? Give me three specific examples."
What to ignore for now: The technical details of how Codex works under the hood. The important insight is not the technology — it is that normal professionals are treating agents as everyday work tools.
2. Samsung Gives ChatGPT and Codex to Every Employee
What happened: Samsung Electronics announced it is rolling out ChatGPT Enterprise and Codex to all employees in South Korea and to its global Device eXperience division. This is one of the largest corporate AI deployments so far.
Why it matters: Samsung previously banned employee use of ChatGPT over data concerns. The fact that they reversed that ban and are now rolling it out company-wide signals two things: the security concerns were solved (ChatGPT Enterprise has stronger data protections), and the productivity gains are large enough to justify the risk. When one of the world's largest companies decides AI tools are standard work equipment, the question shifts from "should I use this?" to "how should I use this responsibly?"
Ask your Ally: "What are the best practices for using AI tools safely at work? What should I never put into a chat prompt?"
What to ignore for now: The specific security certifications or enterprise contract details. The takeaway is that major employers are now actively encouraging AI tool use, not banning it.
3. ChatGPT Comes to KakaoTalk Group Chats
What happened: OpenAI partnered with Kakao, the company behind KakaoTalk — Korea's dominant messaging app used by over 90% of the population. Users can now ask ChatGPT questions and get responses directly inside KakaoTalk group chats.
Why it matters: This is a big deal because it puts AI assistance where people already talk to each other. You do not open a separate app, copy-paste, or switch contexts. You just ask in the group chat and the answer appears alongside your friends' messages. This pattern — AI integrated into existing communication tools — is likely to spread. It makes AI feel less like a separate tool and more like a natural part of how you work and communicate.
Ask your Ally: "How could an AI assistant work inside the apps and tools I already use every day? What would that look like for my workflow?"
What to ignore for now: Which specific countries or platforms this is available on. The pattern is more important than the launch geography.
Practical takeaway for the week: AI agents are becoming standard equipment — not just for developers, but for anyone who does knowledge work. The trend is real. If you have not tried giving your Ally a real task this week, this is the week to start.
Try this this week: Pick one task from Five Things To Try With Your Ally Today and complete it. If you already use AI tools, ask your Ally the follow-up questions listed above. The goal is not to read about AI trends. The goal is to use one.