OpenAI's ChatGPT Work agent connects Gmail, Slack and Drive to complete tasks — what it means for Australian mid-market operators weighing ChatGPT for business.
OpenAI has launched ChatGPT Work, an AI agent designed to connect directly with the workplace apps millions of Australian employees already use every day — Slack, Gmail, Google Drive, Microsoft Teams, Salesforce and calendar tools. The product, reported by SmartCompany on 10 July 2026, marks OpenAI's clearest move yet from a chatbot that answers questions to a system that completes multi-step tasks across the software stack a business already runs on.
For any Australian operator asking what ChatGPT for business actually does differently, the answer is connection rather than conversation. ChatGPT Work can pull information from linked apps, analyse documents, build spreadsheets, presentations, reports and websites, and — through a feature OpenAI calls Scheduled Tasks — repeat jobs automatically, such as summarising Slack conversations, monitoring customer feedback or producing recurring reports. Users still approve the agent's actions as it works, breaking larger projects into smaller steps rather than executing everything unsupervised.
The models behind it were delayed by the US government, not by OpenAI
ChatGPT Work runs on OpenAI's new GPT-5.6 family, which arrives in three variants: Sol, described as the most capable reasoning model; Terra, a general-purpose standard model; and Luna, a faster, lower-cost option for lighter workloads. What is notable is how GPT-5.6 got here. Its public release was delayed after the Trump administration restricted initial access to organisations it had approved, saying additional safety checks were needed before a wider rollout. That is an unusual detail for a workplace productivity launch — a US government screening process sitting between a model and the businesses that want to use it — and it is worth Australian IT and procurement teams noting, given how much of the mid-market's software stack depends on decisions made in Washington and Redmond rather than Canberra.
Anthropic, Google and Microsoft are all building the same thing at once
ChatGPT Work did not appear in isolation. It follows Anthropic's own workplace push with Claude Cowork, launched weeks earlier, and Claude Tag, which lets Anthropic's assistant work across Slack channels. Google and Microsoft have both expanded their agent offerings in recent months as well. OpenAI has folded much of the capability previously housed in its Codex product into ChatGPT Work, and the company says nearly all of its own internal teams — finance, sales and operations — now use it, citing month-end forecasting reduced from days to hours and faster sales proposal preparation as internal examples.
The pattern is an industry-wide bet that the next competitive front in generative AI is not the model itself but how deeply it can be wired into the tools a business already runs — email, chat, file storage, CRM. ChatGPT Work is rolling out first to Pro, Enterprise and Edu subscribers, with Plus and Business users getting access over the following days, according to OpenAI.
What connecting ChatGPT to Gmail and Google Drive actually changes for a mid-market business
For an Australian business with 50 to 500 staff already running Microsoft 365 or Google Workspace, ChatGPT Work's pitch is straightforward: instead of a staff member manually pasting a document into a chatbot, the agent reaches into Gmail, Google Drive or Slack directly, works across several linked systems, and reports back. That is a meaningful shift from how most businesses currently use ChatGPT — as a standalone tab, disconnected from the systems that hold the actual data.
It also raises the exact question that governance teams have been circling since staff first started copying client documents into consumer AI tools. Pacific Data's own view, based on work inside Australian mid-market operations, is that the risk with a workplace AI agent is not the model's intelligence — it is the permissions. An agent connected to Gmail, Google Drive and Slack has, by definition, been granted a level of access most businesses have never formally reviewed for a piece of software, let alone one that acts semi-autonomously across email and file systems overnight.
The arms race is now about permissions, not just capability
OpenAI, Anthropic, Google and Microsoft are each racing to be the agent that sits inside a company's inbox, chat and file storage — because whoever wins that position becomes structurally difficult to remove. Once ChatGPT Work is scheduling tasks, summarising Slack threads and drafting reports on a recurring basis, switching providers means unwinding a genuine operational dependency, not just cancelling a subscription.
What remains unresolved is how Australian regulators and enterprise IT departments will treat an AI agent that has standing access to a company's Google Drive and Slack history, rather than a chatbot a person opens and closes. ChatGPT Work, Claude Cowork and their forthcoming rivals from Google and Microsoft are being rolled out faster than most businesses' governance frameworks were built to handle — and that gap, more than any feature comparison between the platforms, is the story worth watching from here.
