OpenAI Launches ChatGPT Work to Rival Claude Cowork: What UK Businesses Need to Know
OpenAI’s new ChatGPT Work agent takes on Anthropic’s Claude Cowork. Here’s what UK businesses should know before adopting either.
OpenAI just launched an AI agent built to do your job for hours at a time. Not chat with you about it. Do it.
ChatGPT Work went live on 9 July, merging the standard ChatGPT interface, the Codex coding tool and a background automation agent into one desktop app for macOS and Windows. Powered by GPT-5.6, it takes an outcome you describe, gathers context from connected apps and files, breaks the job into steps, and works through them independently — producing finished spreadsheets, slides, documents and even hosted websites rather than chat replies.
UK businesses keep asking me which AI tool to bet on for actual work output, not just conversation. This launch is the clearest answer yet that “finished work” is where the whole industry is heading.
What ChatGPT Work Actually Does
The pitch is straightforward: describe an outcome, not a prompt. Instead of asking ChatGPT to draft a paragraph, you ask it to produce a finished quarterly report, and it goes off, pulls data from connected sources, structures the document, and comes back with something close to done — not a rough draft needing three more rounds of edits.
That’s a genuine shift from how ChatGPT has worked since 2022. The old model was reactive: you asked, it answered, you asked again. ChatGPT Work is closer to delegating a task to a junior analyst and checking back in a few hours. The interface reflects that too — instead of a single scrolling chat log, tasks sit in a queue you can check on, approve, or redirect mid-way through.
How It Compares to Claude Cowork
This isn’t happening in a vacuum. Anthropic launched Claude Cowork months earlier, an agent built around the same idea — planning and executing multi-step tasks with minimal hand-holding. Claude then expanded Cowork to mobile and web this year, letting sessions and files follow users across devices, with background work, scheduled tasks, shared projects and mobile approval prompts.
ChatGPT Work is OpenAI’s direct answer to that. Where Cowork leans on session continuity across devices, ChatGPT Work leans on merging three previously separate OpenAI products — chat, Codex and automation — into a single app. Different architecture, same underlying bet: businesses want agents that finish jobs, not tools that need constant supervision.
I’ve tested both product families over the past year for client work, and the honest answer is neither has fully solved the trust problem yet. Both still need a human checking the output before it goes anywhere important. What’s changed is how convincing the first draft looks — which is exactly why the review step matters more, not less.
A Quick History of How We Got Here
Two years ago, “AI agent” mostly meant a chatbot that could browse the web and summarise a page. Then came coding agents — Claude Code launched into public preview in February 2025 and quietly became one of the fastest-growing products either company has shipped, reaching $1 billion in annualised revenue within its first year.
That success reshaped both companies’ roadmaps. If an agent could reliably write and ship code with minimal supervision, the logical next step was extending that same autonomy to spreadsheets, documents, presentations and eventually entire small websites — which is precisely what both ChatGPT Work and Claude Cowork now attempt.
The Technical Details Worth Knowing
GPT-5.6 sits underneath ChatGPT Work, and OpenAI claims the smaller version of the model performs close to the largest one at roughly a fifth of the running cost. That matters for UK small businesses more than the headline features do — cost per task, not raw capability, decides whether these tools get adopted outside big enterprise IT budgets.
The desktop app also ships a hosted websites feature, letting users build and publish a site directly through ChatGPT Work without touching a separate hosting provider. Combined with the Codex merge, it’s clearly aimed at replacing several small SaaS subscriptions with one interface.
Why OpenAI Is Racing Right Now
Context matters here. Anthropic’s revenue has been climbing hard — run-rate revenue surpassed $30 billion, up from roughly $9 billion at the end of 2025, with business customers spending over $1 million annually more than doubling to exceed 1,000 in under two months. Claude Code alone reportedly hit $2.5 billion in annualised revenue by February, up from $1 billion just a couple of months earlier.
That kind of growth curve puts real pressure on a competitor. OpenAI is reportedly on course for $25-33 billion in annualised revenue for 2026 — still larger in absolute terms, but Anthropic is closing the gap fast on the enterprise coding and agent side specifically, which is exactly where ChatGPT Work is aimed.
When I looked into the timing, launching an enterprise-focused agent product right as a rival’s enterprise revenue triples isn’t a coincidence. It’s a company protecting its most profitable segment before a competitor locks in more of it.
What This Means for the Wider AI Race
Google isn’t sitting this one out either. Anthropic recently announced an expanded partnership with Google and Broadcom for multiple gigawatts of next-generation compute — its most significant infrastructure commitment to date. That’s not a product announcement, but it signals Anthropic is scaling hard enough that it needs Google’s chip supply chain to keep up with demand.
Three companies, three different bets: OpenAI merging products into one agent app, Anthropic scaling infrastructure and expanding Cowork across devices, Google supplying the compute underneath both while building its own Gemini agents in parallel. UK businesses choosing a vendor this year are picking a horse in a race that’s still very much open.
The Practical Risk for UK Adopters
Here’s what doesn’t get said enough in the coverage: agent tools that produce “finished work” still need review. A spreadsheet that looks complete and a spreadsheet that’s actually correct are not the same thing, and the failure mode with autonomous agents is that mistakes look more polished than they used to.
UK businesses keep asking about liability too — if an agent drafts a contract clause or a financial report with an error, who’s accountable? Neither OpenAI nor Anthropic has published clear guidance on this for UK-specific regulatory contexts, and that gap matters more as these tools move from chat assistants to task-completing agents with real output.
I’d treat both ChatGPT Work and Claude Cowork the same way right now: useful for cutting first-draft time dramatically, not yet trustworthy enough to skip human review on anything client-facing or financial. Save the biggest time gains for internal, low-stakes work until the track record improves.
How This Affects Freelancers and Small Agencies
Solo operators and small UK agencies stand to gain the most from tools like this, at least on paper. A one-person marketing consultancy that used to spend a day building a client report can now generate a first pass in an hour and spend the rest of the day on strategy and client conversations instead.
The catch is the same one enterprise buyers face, just with less margin for error. A small agency doesn’t have a compliance team to catch a mistake before it reaches a client. That makes the human review step even more important for freelancers than for a large firm with layers of sign-off already built into its process.
Pricing and Availability
OpenAI hasn’t published detailed UK pricing tiers for ChatGPT Work at launch, though the desktop app is rolling out for macOS and Windows users on existing paid plans first, with wider availability expected to follow. Claude Cowork, by comparison, is available to Claude’s existing Team and Enterprise subscribers, with mobile access added this year.
Neither company has confirmed UK-specific data residency terms for the new agent features, which UK businesses in regulated sectors — finance, healthcare, legal — will want clarity on before connecting sensitive internal systems to either tool. That’s not a small caveat. A firm handling client financial data needs to know exactly where that data sits and who can access it before an agent starts pulling from internal folders automatically.
What Early Adopters Are Reporting
Coverage from launch day describes ChatGPT Work staying on complex projects “for hours,” a phrase OpenAI itself used in briefings. Early testers cited in press coverage described the output quality as a genuine step up from previous chat-based drafts, though several noted the same caveat any agent tool earns early on: it’s excellent at structure and formatting, less reliable on facts it hasn’t been explicitly given.
That’s a pattern worth remembering with any agent, from either company. Structure and polish come easily to these models. Accuracy on specifics still depends entirely on what data you feed them, and neither tool is any better than its source material.
What This Means for UK Businesses
If your team already uses ChatGPT or Claude day to day, this is worth testing on low-stakes work first: draft reports, internal documents, first-pass website builds. Don’t hand either agent anything client-facing or regulatory without a human check at the end, at least for now.
The bigger picture matters more than either single launch. Agent-based AI that finishes tasks rather than just answering questions is now the default direction for both major providers, and UK businesses that wait another year to experiment will be catching up rather than learning alongside everyone else.
Stay ahead of the market
Join our community of nearly 5,000 across YouTube, LinkedIn, X, and Facebook — weekly crypto, AI, and digital lifestyle insights every Thursday. No spam. Unsubscribe any time.
Partner picks
Build a smarter digital stack
Explore curated AI, automation, wealth, and creator tools selected for practical value, transparent pricing, and clear use cases.
Disclosure: some links may be affiliate links. DigitechLifestyle may earn a commission at no additional cost to you.
