Services AI consultingAI implementationTeam trainingLectures MethodInsightsAboutContact Book a call
‹‹ Field notes

Handing the weekly status update to a ChatGPT Business team task

ChatGPT Business can now run a recurring team task, answer in Slack or Teams as @ChatGPT, and power internal Sites with live data. How to move the weekly status update onto it, what it costs in seats and credits, and what to test first.

Updated October 5, 2026: rewritten as a full setup guide, with corrected details on Sites, the $50 credit and team ownership.

In most small companies the weekly status update depends on one person: they collect it, write it, and chase whoever forgot to report. On September 29 OpenAI gave ChatGPT Business a way to move that job off one desk. Team tasks run recurring work and post the result where the team reads it, @ChatGPT answers questions inside Slack and Microsoft Teams, and Sites can now pull live data from connected apps.

What changed

Team tasks run in the cloud on a schedule, on a set date, or on an event such as a new Gmail email or Slack message. They use company connections an admin approved, and any teammate with permission can view, run, edit or pause them. OpenAI's own example is a weekly project update posted to a Slack channel, with progress, blockers, next steps and source links.

@ChatGPT works inside Slack and Microsoft Teams once an admin installs it. People mention it in a channel or message it directly, and it answers from the company tools connected for it.

Sites, the internal pages Business members build by describing them, has been in public beta since July. On September 29 it gained connected apps, editing and schedules.

What it means for a small business

Picture a marketing agency of 15. Every Monday the operations lead spends the morning reading Slack threads, a deadlines sheet in Google Drive and the client folders, then writes a client-by-client update for the team channel. As a team task, the update runs at 8:00 on Monday, reads the same sources through the company's connections and posts a draft to the channel with links. The operations lead reviews it instead of writing it, and when a client is added, anyone on the team with permission edits the instructions.

The bigger change is where the knowledge lives. Today the update's logic sits in one person's habits. In a team task it sits in written instructions the whole team can read and fix, which is what OpenAI means when it says maintaining the task "doesn't depend on one person."

Two things need care. The task runs under the team's service account, through connections whose access can exceed any one member's; OpenAI warns that a workspace connection "may read sources you cannot open." And every run draws on usage the workspace pays for.

How to set it up

  1. Get the plan and permissions. Business needs at least two seats. The workspace owner gives the right people permission to create teams and team tasks.
  2. Create the connections. In the Admin Console, open Workspace connections, click +, pick the app and sign in with a dedicated company account, not someone's personal one. OpenAI documents this for Google Drive, Gmail, Slack, Notion, GitHub and Linear. Grant the connection to the team.
  3. Create the team. Settings, then Teams, then Create team. The creator becomes the single owner.
  4. Write the task. In the team's Scheduled section, start a new task. The instructions name the sources, the format (progress, blockers, next steps, links), the destination channel and what to flag. Choose a weekly schedule and check the time zone.
  5. Run it by hand twice. Use Run now, read the result in Previous runs, check the channel, and refine. A "Completed" schedule only means no runs are left; it doesn't confirm the work succeeded.
  6. Add @ChatGPT for follow-ups. In the Admin console, open Agents, add Slack or Teams, create a surface, limit it to selected channels and add @ChatGPT to the status channel.
  7. Cap the spend. Under Workspace settings, then Billing, set a monthly recharge limit and per-user credit limits, and give @ChatGPT a spend cap.

Costs, limits and risks

Business costs $25 per user per month, or $20 billed annually, in most countries, with a two-seat minimum; prices vary by country and currency. A 15-person agency on annual seats pays $300 a month. Seats include some usage; beyond it, team tasks and @ChatGPT draw on workspace credits, and OpenAI doesn't publish what one run costs. Purchased credits last 12 months, and automatic reload has no ceiling unless you set a monthly recharge limit. Since September 8 there's also a $50 credit offer for connecting a partner plugin you haven't used before, like QuickBooks or HubSpot, and completing one workflow within seven days.

As a rough estimate: if the update takes three hours to compile and 30 minutes to review, that's about ten hours a month back. Treat it as arithmetic on assumptions, not a promise.

The risks to manage:

Should you do it now?

Do it now if you're on Business, the update draws on Google Drive, Gmail or Slack, and one person spends hours on it every week.

Wait if your status lives in HubSpot or Microsoft 365, or the update has to post in Teams, until a test shows the connection can read and post there. Leave Sites for later: connecting it to company apps needs two admin switches that are off by default, and it's still a beta with pricing to come.

Skip it if you work alone. Plus already includes personal scheduled tasks, which cover a one-person update.

What I'd do this week

What I'd do this week: write down where the Monday update's information comes from and where it should land. Connect those sources with a dedicated company account, build the team task, and run it twice by hand before Monday. Success is an update the team reads without asking anyone what it means.

Want this running inside your business?

A 20-minute conversation. No pitch, no pressure. We look at where AI would actually save your team time.

‹‹ All field notes