OpenAI just turned its chatbot into a digital employee that clocks in when you clock out.
The Summary
- OpenAI launched ChatGPT Work, an AI agent powered by GPT-5.6 that executes multi-step tasks across email, Slack, calendars, and code repositories without human intervention
- GPT-5.6 received Trump administration approval for public rollout after a two-week "limited preview" period restricted to government-approved organizations
- ChatGPT Work represents OpenAI's pivot from Q&A tool to autonomous work platform, launched as the company filed for an IPO at valuations between $730-852 billion
- The timing is strategic: OpenAI is packaging agent capabilities from its internal Codex tool for non-technical users just as it positions itself for what could be the largest tech IPO ever
The Signal
OpenAI filed its draft S-1 with the SEC last month with annualized revenue past $25 billion. Now it needs to show public market investors that ChatGPT is more than a feature. ChatGPT Work is that story. It's the productization of agent economics: software that doesn't wait for prompts, but executes standing orders.
The regulatory path matters here. GPT-5.6 spent two weeks in a government-only preview, a detail both sources flag but don't explain. That's not normal product rollout cadence. That's the new reality of frontier AI deployment: you show the government first, get approval, then ship. The Trump administration greenlit it. Now ChatGPT Work runs on what Sam Altman calls "the best model we have ever produced", a suite of three variants called Sol, Terra, and Luna.
"ChatGPT Work can gather context from connected apps, files, and workflows to produce finished documents, spreadsheets, presentations, reports, and websites."
What's notable is the durability claim. The agent stays with complex projects for hours, completing them independently. Not "helps you draft." Not "suggests next steps." Completes. That's the language of replacement, not augmentation. OpenAI's pitch is that it's democratizing Codex, the internal engineering tool that already runs autonomous workflows for OpenAI's own teams. Ty Geri, the product manager who built ChatGPT Work, framed it as bringing "agentic AI capabilities" to everyday users.
The integration surface is broad:
- Email inboxes and outboxes
- Slack channels and DMs
- Calendar scheduling and meeting prep
- Code repositories for technical users
- Document and spreadsheet generation
This isn't a single-app add-on. It's a layer that sits across your work stack. The value prop is ambient delegation: you give it an outcome, it figures out the steps, connects to the tools, and ships the result. That's fundamentally different product design than "chat with your documents." It's closer to hiring a junior analyst who never sleeps and costs $20/month instead of $60K/year.
The Implication
If ChatGPT Work delivers on the hours-long autonomous task execution claim, knowledge work pricing is about to get weird. The wedge is administrative and coordination work, the stuff that fills calendars and generates status update emails. Once agents prove they can own those workflows end-to-end, companies will staff differently. Not in five years. In the next budget cycle.
For individuals, the question is whether you're building skills that compound or tasks that agents will eat. If your job is assembling information other people produced into formats other people expect, you're in the blast radius. If you're generating the insights agents assemble, you just got a force multiplier. Watch how fast "prompt engineering" stops being a job title and starts being table stakes.