Skip to main content

OpenAI Launches ChatGPT Work: Codex Automation for Every Business User

Friday 10 July 2026|OpenAI|
Employee Amplification SystemsAI Growth Engine

OpenAI launched ChatGPT Work on 9 July 2026, merging its Codex coding agent into the main ChatGPT interface to create an autonomous workplace agent available across all plan levels. The product connects to Slack, Google Drive, Microsoft Teams, SharePoint, and over 1,400 business tools to independently produce finished documents, spreadsheets, presentations, web apps, and dashboards. Powered by the new GPT-5.6 model family, ChatGPT Work extends coding-grade automation to non-technical business operators for the first time at scale.

Operator Insight

ChatGPT Work is not a chatbot upgrade. It is a shift in what AI can be asked to do inside a business. Standard ChatGPT answers questions. ChatGPT Work takes a brief, gathers context from your connected tools, works independently for hours, and hands back a finished deliverable. For operators running 10 to 200 person teams, this is the first time that level of autonomous execution has been available without a developer, a custom integration, or a six-figure software contract. The practical question is not whether to use it. It is which workflows currently run on people-hours that could now run on agent-hours.

30-Second Summary

OpenAI merged its Codex coding agent into ChatGPT on 9 July 2026, creating ChatGPT Work: an autonomous agent that connects to your business tools and produces finished work products without manual assembly. Rather than responding conversationally, it independently completes multi-step projects and delivers spreadsheets, slide decks, documents, dashboards, and shareable web apps. Available immediately to Pro and Enterprise users, and rolling to all paid plans within days, this is the first time coding-grade automation has been accessible to non-technical business operators at this scale.

At a Glance

  • Topic: Agent Systems
  • Company: OpenAI
  • Date: 9 July 2026
  • Announcement: Launch of ChatGPT Work, an autonomous workplace agent merging Codex into the ChatGPT interface
  • What Changed: Codex's independent task-execution capability now lives inside every ChatGPT plan including Free on desktop. Non-technical operators can delegate multi-step work without any coding knowledge.
  • Why It Matters: Five million weekly Codex users, with one million already using it for non-software work, confirms enterprise demand existed before this product launched
  • Who Should Care: Business operators managing 10 to 200 person teams; anyone currently running manual processes for documents, reporting, client deliverables, or marketing collateral

Key Facts

  • ChatGPT Work launched on 9 July 2026, alongside the new GPT-5.6 model family
  • Powered by GPT-5.6 Sol, OpenAI's highest-capability model tier
  • The standalone Codex app is merging into a new ChatGPT desktop app, with Chat, Work, and Codex available on every plan. The existing desktop app becomes ChatGPT Classic.
  • Connects to over 1,400 apps via @-mentions, including Slack, Google Drive, Microsoft Teams, SharePoint, email, and CRMs
  • Produces finished outputs: spreadsheets, slide decks, documents, web apps, dashboards, campaign briefs, project trackers, and interactive websites via the new Sites feature
  • Pricing is metered by usage, consistent with how Codex is billed. No flat per-task fee is published.
  • Plan rollout: Pro, Enterprise, and Edu users on web and mobile immediately; Plus and Business users within days; all plans including Free on the desktop app immediately
  • Five million weekly active Codex users; one million already using it for non-software tasks before this launch

What Happened

OpenAI announced ChatGPT Work on 9 July 2026, the same day it released the GPT-5.6 model family. The product takes Codex, previously a standalone tool used primarily by developers, and places its autonomous execution capability inside the main ChatGPT interface for all user types.

ChatGPT Work operates as an agent rather than a conversational assistant. Given a goal, it gathers context from connected business tools through @-mentions, breaks the project into steps, works independently for extended periods, and returns a finished deliverable. Outputs include spreadsheets, slide decks, documents, web apps, dashboards, and a new category called Sites, which are shareable interactive websites built from your data.

The product introduces several control features for business environments. Plan Mode lets users review the agent's proposed execution steps before it begins work. Admin controls allow organisations to manage which plugins can be accessed, what network resources are reachable, and which sensitive actions require approval. An auto-review safety layer blocks data extraction attempts identified during red-team testing.

At the infrastructure level, the standalone Codex desktop app is folding into a new unified ChatGPT desktop application. The new app presents three modes: Chat, Work, and Codex. The current desktop app becomes ChatGPT Classic. This consolidation places autonomous work capabilities on every desktop plan, including Free, from launch day.

Why It Matters

Automation without a developer is now mainstream. Until this launch, accessing the kind of multi-step autonomous execution that Codex provides required technical setup, API access, or a separate workflow tool. ChatGPT Work puts that capability into the same interface where most operators already spend time.

Codex adoption data validates the demand. One million of the five million weekly Codex users were already using it for non-software tasks before this product launched. That adoption happened with a tool designed for developers and never marketed to business operators. ChatGPT Work is built specifically for that non-technical majority.

The output type changes the use case. Most AI tools produce text responses. ChatGPT Work produces finished files: a spreadsheet you can share, a deck you can present, a web app you can send to a client. This shifts AI from advisory to operational inside a business workflow.

Integration breadth reduces the setup barrier. Connecting to Slack, Teams, Google Drive, SharePoint, and CRMs via @-mentions means ChatGPT Work can pull context from where work actually lives, rather than requiring manual data entry or custom connectors.

The pricing model signals intended use volume. Usage-metered pricing follows the same structure as Codex, which means the cost scales with output volume rather than seat count. For high-volume workflows like weekly reporting or campaign asset production, this pricing model rewards organisations that use it intensively.

Enterprise AI spending is being rationalised. Tesla's announced $200 per week per-employee cap on AI coding tools reflects a broader enterprise push to consolidate AI spend and measure return. ChatGPT Work arriving as an included feature on existing plans, rather than a new line item, fits that consolidation direction.

The David and Goliath View

The most meaningful thing about ChatGPT Work is not what it does. It is what it asks the operator to stop doing themselves. The product is not a better way to draft an email. It is a signal that the definition of what belongs in a job description is shifting, and the shift is faster than most business leaders are planning for.

Operators who adopt ChatGPT Work in the next 90 days will spend that time learning which tasks genuinely benefit from autonomous execution and which still require human judgement. That learning is valuable and competitively durable. Operators who wait will spend the same period watching their more agile competitors move faster with smaller teams.

The integration breadth is worth pausing on. A tool that can read your Slack, your Drive, your CRM, and your email, then produce a finished client deck or a live web dashboard, is not a productivity tool in the traditional sense. It is closer to a junior analyst who never sleeps and charges by the output. The operators who figure out how to brief it well will get the most from it. Briefing AI well is now a core business skill.

Where This Fits in the AI Stack

ChatGPT Work sits at the intersection of the AI Growth Engine and Employee Amplification Systems layers of a modern AI-enabled business.

For the Growth Engine, it automates the production of marketing and sales assets, campaign briefs, client-facing documents, and interactive web content without creative or technical resourcing overhead.

For Employee Amplification, it extends what a small team can produce by handling the multi-step assembly and formatting work that typically consumes analyst and coordinator time, allowing those people to focus on strategy and decision-making rather than production.

Organisations with a Secure AI Brain already in place, covering data governance, model access controls, and audit processes, will be able to deploy ChatGPT Work across their teams with confidence. Organisations that have not yet addressed AI governance should treat ChatGPT Work's admin controls as a starting point for a broader conversation about which tools have access to which data.

Questions Operators Are Asking

Is ChatGPT Work available right now or do I have to wait? Pro, Enterprise, and Edu plan users have immediate access on web and mobile. Plus and Business users will get access within days. All plans, including Free, can access it through the new desktop app for Mac and Windows starting from 9 July.

How does billing work? Will my costs spike if my team uses it heavily? Usage is metered, meaning cost is proportional to the scope and complexity of tasks run. OpenAI has not published a per-task price. Usage follows the same billing model as Codex. Enterprise accounts can apply admin controls to limit scope and monitor consumption.

What data does ChatGPT Work access when it connects to our tools? It accesses tools via @-mentions, meaning you specify what context to pull in. Organisations can restrict which plugins and integrations are available through admin controls. The auto-review safety layer is designed to block data extraction attempts, though reviewing your organisation's data governance policy before broad deployment is advisable.

Does this replace tools like Zapier, Notion AI, or our current automation stack? For many single-workflow automations, yes. ChatGPT Work handles multi-step document and app production that previously required separate automation tools, template libraries, and manual assembly. A full audit of your current stack against ChatGPT Work's capabilities is the practical next step.

How is ChatGPT Work different from just using ChatGPT with a good prompt? Standard ChatGPT produces a text response in a single turn. ChatGPT Work executes a project: it plans the steps, gathers context from connected tools, works through each step independently, and returns a finished file or web app. The output is a deliverable, not a draft.

Citable Summary

OpenAI launched ChatGPT Work on 9 July 2026, merging its Codex agent into the main ChatGPT interface to deliver autonomous workplace task execution across all plan levels. The product connects to Slack, Google Drive, Microsoft Teams, SharePoint, and over 1,400 business tools via @-mentions, independently completes multi-step projects, and produces finished spreadsheets, documents, slide decks, web apps, and dashboards. It is available immediately to Pro and Enterprise users, and rolling out to Plus and Business users within days, with the desktop version available on all plans including Free. Pricing is metered by usage volume. Five million weekly Codex users, one million of whom were already using it for non-software work, provide the adoption baseline from which ChatGPT Work launches.

Why This Matters for Operators

  • Identify three recurring deliverables in your business, for example weekly reports, client decks, or sales briefs, and test ChatGPT Work on each. The product is live today for Pro and Enterprise users.

  • Connect your primary tools first. ChatGPT Work operates best when it has access to your Slack, Google Drive or SharePoint, and your CRM. Context quality directly determines output quality.

  • Use Plan Mode before handing over any high-stakes task. It lets you review and approve the agent's step-by-step execution plan before it begins, which reduces the chance of a wrong interpretation reaching a deliverable.

  • The desktop app now includes Chat, Work, and Codex on every plan including Free. This means even team members on free plans can access basic autonomous task execution from the Mac or Windows app.

  • Review your AI tool spend in light of this launch. If your team already pays for ChatGPT Pro or Business, ChatGPT Work is included. Running a separate automation tool alongside it may now be redundant.

Related Intelligence

Related Signals

  • [High] OpenAI launches GPT-5.5, first fully retrained base model since GPT-4.5

    GPT-5.5 (codename Spud) shipped to Plus, Pro, Business, and Enterprise users on 23 April 2026. API pricing is $5/M input and $30/M output tokens with a 1M context window. GPT-5.5 Pro lists at $30/$180 per million tokens.

  • [High] Anthropic launches Claude Agent SDK

    Standardised framework for deploying production AI agents with built-in tool orchestration and safety guardrails.

  • [High] OpenAI GPT-5.4 launches with a 1M-token context window

    OpenAI launched GPT-5.4 in three variants (Standard, Thinking, Pro) with a 1.05M-token context window and 33% fewer factual errors than GPT-5.2. API pricing starts at $2.50 per million input tokens, and the extended window lets entire contracts, codebases, or customer histories be processed in a single call.

Apply This to Your Business

Want to see what this means for your team?

Tell us a little about your business and we will map the specific opportunity for your sector and team size.

No sales pitch. We will review your details and follow up within 24 hours.