July 9, 2026: The day two rival AI labs accidentally proved they’re terrified of each other by shipping nearly identical products within hours of one another.
You know that moment in a spy movie where two undercover agents realize they’re after the same target at the same party, and they just stare at each other across the ballroom? Anthropic and OpenAI had that moment last Tuesday—except instead of a tuxedo, they were both wearing “agentic AI for knowledge workers,” and instead of gunfire, they used press releases.
The Setup: Agentic AI Goes Mainstream
Here’s what happened: Anthropic had been quietly flexing Claude Cowork since January—a desktop agent that could read your emails, crack open spreadsheets, draft presentations, and execute multi-step workflows without you basically narrating every move like you’re directing a hostage negotiation. It was good. Builders loved it. But it only worked on your laptop.
Then, on July 9, Anthropic expanded Cowork to web and mobile. Suddenly, you could start a 3-hour research task on your MacBook at 9 AM, hand it off to your phone at lunch, and check the results from a meeting by 3 PM—all while Claude kept working in the background like the world’s most competent intern.
Same day. Same hour. OpenAI launched ChatGPT Work.
Complete. Identical. Vibes.
The Unspoken War
This wasn’t a coincidence. Both companies are racing toward IPOs worth $800B+ (Anthropic’s valued at ~$965B, OpenAI confidentially filing at ~$852B). Institutional investors don’t want to own just another model company—they want to own the interface where knowledge workers actually use AI every day. That’s where the recurring revenue, the stickiness, and the defensibility live.
So Anthropic went first with Cowork’s mobile expansion. OpenAI, instead of building differently or slower, essentially said: “Same thing. But with GPT-5.6. And cheaper.”
GPT-5.6’s three-tier pricing (Sol, Terra, Luna) undercuts Anthropic’s Claude Fable 5 by 50% or more at every capability level. Plus, ChatGPT already has hundreds of millions of users. Claude Cowork is still rolling out to Max subscribers. Do the math.
What’s Actually Winning
Here’s the thing nobody’s talking about loud enough: 90% of Claude Cowork’s sessions have nothing to do with coding.
Ninety percent. That’s not “engineers using an AI assistant to write functions.” That’s researchers drafting literature reviews, sales teams pulling data from five CRMs, finance teams reconciling spreadsheets, product managers prepping decks. Anthropic pulled this data from 1.2 million real sessions. The narrative everyone was told—“AI agents for developers”—turned out to be wrong the entire time.
So now both companies are basically saying: “Actually, AI agents are for everyone.” And both are willing to price-war their way into your workflow.
The pivot is stunning: twelve months ago, the conversation was “How big is your model?” Now it’s “How much does a finished task cost?” Raw intelligence stopped mattering the second two companies both claimed to have solved “good enough.” Now it’s pure ruthlessness on price and UX.
What the Internet Is Saying
Of course, the tech world had thoughts:
Sam Altman (OpenAI CEO): “ChatGPT Work is here. Agents that work for you. Running complex multi-step tasks, integrating with your tools, finishing what they start. No more hand-holding required.” Happy building 🎉
Builder feedback (split): Some devs report GPT-5.6 agents are more reliable for routine work. Others swear Anthropic’s models still win on raw reasoning. The real dividing line isn’t intelligence anymore—it’s whether you trust the cost predictability.
Enterprise finance teams (via Reddit/HN): “Finally. We were getting buried in AI token bills. Price war is the only thing that makes sense right now.”
The Hot Take
The company that wins this isn’t the one with the “smarter” model. It’s the one that makes knowledge workers feel like they have a ghost coworker who just… does stuff when they ask, costs roughly what a mid-tier SaaS tool costs, and doesn’t make you second-guess whether you should be using it.
Anthropic got there first. OpenAI’s playing catch-up with a bigger install base and a price gun.
Your team’s productivity is about to get weird.
This post has been created by Claude AI.
References
- Claude Cowork on web and mobile: hand off work anywhere — Anthropic Blog
- OpenAI unveils long-awaited ‘super app’ as rivalry with Anthropic intensifies — CGTN
- OpenAI’s ChatGPT Work Takes Aim at Claude Cowork — Technology.org
- Claude Cowork Goes Mobile: AI Agents Work While You Sleep — Enterprise DNA