ChatGPT Workspace Agents: Codex Team Automation Guide (2026)

On April 22, 2026, OpenAI launched workspace agents in ChatGPT — and the difference between this and previous GPTs is not cosmetic. Where GPTs were essentially saved system prompts with a handful of connected tools, workspace agents are cloud-running processes that can be scheduled, deployed into Slack, write and execute code, remember context across sessions, and operate on your team’s data without you being present. If you have been waiting for ChatGPT to feel less like a chatbot and more like a capable teammate, this is the release that changes that calculus.

This guide covers everything your team needs to know: what workspace agents actually are, how they differ from GPTs, how to set one up in under ten minutes, which integrations matter most, and what the pricing looks like after the free preview ends on May 6.

What Are Workspace Agents? (And Why They Are Not Just Better GPTs)

Workspace agents are OpenAI’s evolution of GPTs — but calling them “better GPTs” undersells the architectural shift. GPTs were conversational templates: you gave ChatGPT a system prompt, maybe connected a few tools like web browsing or DALL·E, and saved it for reuse. The interaction model was synchronous — you prompt, it responds, done.

Workspace agents are built for asynchronous, multi-step work. They run in the cloud so they keep working after you close the browser. They can be triggered on a schedule. They can receive requests from Slack and respond autonomously. They retain memory across sessions — what they learned from last week’s run informs this week’s. And critically, they are shared at the workspace level, meaning a team can build one agent and every member benefits from it.

The underlying engine is Codex — the same model powering OpenAI’s developer coding assistant — which means workspace agents can not only plan and reason about multi-step tasks but actively write and run code as part of completing them. An agent tasked with preparing a weekly analytics summary can query your connected data sources, write a Python script to process the output, and deliver a formatted report, all without your intervention.

How Workspace Agents Work: The Architecture

Understanding the architecture helps you build better agents. Each workspace agent consists of three layers working together.

Instructions and Persistent Memory

The agent’s core identity: what it does, what tone it takes, what it should and should not attempt. This is similar to a GPT system prompt but informed by a persistent memory layer — the agent updates its own understanding of your workflows over time. If you correct its report format in week two, it remembers that adjustment in week three without you re-prompting.

Connected Tools and Apps

Workspace agents connect to over 60 apps directly from ChatGPT Business workspaces, including Slack, Google Drive, SharePoint, GitHub, Jira, Atlassian, Notion, Microsoft 365, and Salesforce. Each integration grants the agent specific permissions — read-only versus read-write, specific channels or folders — that workspace admins configure through the ChatGPT admin console.

Beyond pre-built integrations, agents can write and execute code (primarily Python), browse the web, and call external APIs where webhooks are configured. This means an agent can pull data from a proprietary internal system through an API call, process it with a Python script, and push the result to a Slack channel — in a single autonomous run.

Triggers

Three trigger modes are available at launch:

• Manual: A team member opens the agent in ChatGPT or Slack and asks it to do something.
• Scheduled: The agent runs automatically at configured intervals — daily, weekly, on the first Monday of the month. Ideal for recurring reports, status digests, and periodic data pulls.
• Slack-triggered: The agent is deployed into a Slack channel or DM and picks up requests as they arrive, responding without requiring a browser session from any team member.

Getting Started: Set Up Your First Workspace Agent in 10 Minutes

Setting up a workspace agent takes under ten minutes for straightforward use cases. Here is the exact path to follow:

1. Open ChatGPT and click “Agents” in the left sidebar. This is available on Business, Enterprise, Edu, and Teachers plans. If you do not see it yet, the feature is rolling out progressively through April 2026.
2. Describe the workflow you want to automate. Be specific: “I want an agent that every Monday morning pulls the previous week’s GitHub Issues closed by our team, summarizes them into a changelog entry, and posts it to our #releases Slack channel.” ChatGPT will walk you through configuration step by step.
3. Connect the required apps. For the GitHub + Slack example, authorize both integrations with the appropriate permissions: GitHub read access on your repo and Slack post permission to the target channel.
4. Set the trigger. Choose “Schedule” and set it to Monday at 9:00 AM in your workspace’s timezone.
5. Test the agent manually first. Run it once and review the output before enabling the schedule. Iterate on the instructions if the format is off or the scope is too broad.
6. Share with your team. Once you are satisfied with the output, publish the agent to your workspace. It appears in every team member’s Agents sidebar and can be invoked directly from Slack.

High-Impact Use Cases for Teams

OpenAI’s framing is that workspace agents are best suited for “work your team does repeatedly.” Early business feedback points to five categories where the value is clearest.

Weekly Reports and Digests

Any report requiring data from multiple sources, consistent formatting, and scheduled distribution is a strong workspace agent candidate. Sprint summaries from Jira, sales pipeline digests from Salesforce, content performance reports from Google Analytics — the agent handles the data pull, synthesis, and distribution. The team just reads the output.

The multi-agent orchestration capability matters here: the agent can spawn parallel subagents to research different data sources simultaneously. A prompt like “prepare this week’s competitive intelligence digest” triggers parallel research across multiple sources, synthesized into a single coherent report — the kind of work that previously took a junior analyst two hours every Friday morning.

Code Review Assistance

An agent connected to GitHub can monitor pull requests, run the codebase through a Codex-powered review pass, and post structured comments on each PR — flagging missing tests, potential edge cases, or style guide violations. Engineers still own the final review, but the mechanical first pass is automated. Teams using this pattern report that the real value is not catching bugs (reviewers still do that) but standardizing the review checklist so nothing gets forgotten under deadline pressure.

Slack Message Triage

For support teams and community managers, a workspace agent deployed into a Slack channel can triage incoming messages, draft responses for common questions, escalate edge cases to the right person, and maintain a log of resolved issues. This is one of the clearest automation targets for growing teams — the first-line response work that consumes significant hours without requiring deep expertise.

Research and Competitive Intelligence

The parallel subagent capability makes workspace agents significantly more useful for research tasks than a standard ChatGPT session. A prompt like “give me a competitive analysis of our top three rivals’ pricing pages” can trigger simultaneous research across all three, with the results synthesized into a structured output. Compared to running three separate research sessions manually, the wall-clock time drops from 45 minutes to under five.

Document and Proposal Drafting

An agent with access to your Google Drive and SharePoint can draft proposals, status reports, and client communications using your company’s existing documents as context. Unlike a one-off ChatGPT prompt, the agent retains context across sessions: previous drafts, brand voice guidelines, and the feedback you have given over time all inform the next output without re-prompting.

Pricing and Availability

Workspace agents launched on April 22, 2026, as a research preview, with features continuing to expand. Here is the current access picture:

• Available now: ChatGPT Business ($25/month per user billed monthly, $20/month billed annually), Enterprise, Edu, and Teachers plans.
• Not yet available: Free, Go, and Plus plans. OpenAI has not indicated a timeline for lower-tier access.
• Free preview window: Workspace agent usage is free through May 6, 2026. After that date, credit-based pricing applies. Specific rates have not yet been published.
• Launch credits: Eligible Business workspaces can earn up to $500 in credits when team members begin using Codex, which offsets the first months of usage after the free window closes.

The move to credit-based pricing is a meaningful departure from fixed-seat models. Light users consume fewer credits while heavy automation users consume more — likely better for teams with uneven usage patterns, but worth monitoring to avoid unexpected billing after the free window closes.

Current Limitations to Know Before You Build

Workspace agents are a research preview, and the limitations are real. Build your automation strategy around what is actually available today, not the implied roadmap:

• Text-only agent output: While Codex can generate and run code, the agent response surface is text. Image generation and rich media outputs inside workspace agent runs are not yet supported.
• Surface coverage: Currently limited to ChatGPT and Slack. More surfaces, including the Codex app, are on the roadmap but not yet available.
• Admin setup required: App integrations must be authorized by workspace admins. In larger organizations, IT approval cycles can add days or weeks to rollout.
• No public API: There is no way yet to trigger workspace agents programmatically from external systems. Scheduled triggers and Slack are the only automated entry points. An API is expected post-preview but has not been announced with a date.
• Shared memory scope: Agent memory is workspace-wide, not personalized per user. Useful for building shared institutional knowledge, but individual preferences are not tracked separately.

What to Watch in the Coming Weeks

OpenAI’s stated roadmap includes new automated triggers beyond schedules and Slack, performance dashboards for tracking and improving agent output, broader app coverage, and support in the Codex app. The credit pricing announcement on or before May 6 will be the most important data point for teams evaluating workspace agents against dedicated automation platforms like Zapier, Make, and n8n.

The signal to watch most closely is whether OpenAI publishes a public API for workspace agents. If they do, the integration surface expands dramatically and enterprise buyers gain a much cleaner case for standardizing on workspace agents over custom automation stacks. If the API stays closed, workspace agents will remain a strong but bounded tool rather than a platform.

Making the Most of the Free Preview Window

The free preview period through May 6 is an unusually generous sandbox. Three things worth doing before that date:

1. Identify your highest-frequency recurring tasks. The ROI on workspace agents scales with repetition. A task done once a month is a marginal win. A task done daily or weekly that takes 45 minutes each time is a strong automation candidate — and the most likely to generate clear before/after data for a business case.
2. Get your integrations authorized early. Admin approval for GitHub, Slack, and Google Drive can take longer than expected in enterprise environments. Start the IT request now so you are not blocked when you want to build.
3. Build feedback loops into every agent. The best workspace agents include a step that posts a summary of what they did and flags anything that looks anomalous for human review. Fully dark automation — where no human checks the output — tends to drift silently as inputs change over time.

ChatGPT workspace agents are the most concrete implementation of “AI as a team member” that any major AI lab has shipped at scale. The architecture is right, the integrations cover the core enterprise stack, and the free preview gives teams a no-risk window to build real workflows. Watch the credit pricing announcement on May 6 and the API roadmap — those two variables will determine whether workspace agents become the automation layer for teams or remain a useful but bounded productivity tool.