ChatGPT Dots in Slack: Set Up Monitoring, Updates, and Sharing Rules

Put your ChatGPT dot to work in Slack: pick the right connection, set monitoring and schedules, and keep posts and private info behind clear rules.

What Changes When Your Dot Joins Slack

You already know what a ChatGPT dot is and why your team might use one. The ChatGPT dots implementation playbook covered responsibility briefs, connect-only-what-you-need principles, and a generic rollout timeline. This post is different. It walks through the mechanics of putting your dot to work in Slack: how the two connection methods differ, who can direct the dot and who can see its posts, how to set up monitoring and recurring schedules, how to write sharing rules so your dot checks before posting private information in a team channel, and how to review and stop its work when needed.

Dots are rolling out gradually to eligible Pro, Business Premium, and Enterprise plans, and in Enterprise workspaces they stay off until a workspace admin enables them. Check OpenAI's current documentation for the latest availability and settings. The facts in this guide are grounded in OpenAI's admin and user documentation as of October 2026.

Two Ways to Connect Slack

Your dot can work with Slack in two ways, and the distinction matters for visibility and control.

Connect the Member's Slack Account Through Plugins

In this setup, members talk to their dot inside ChatGPT. The dot can use their Slack connection without having its own Slack identity. Configuration happens in two places. First, the member connects their Slack account as a plugin in ChatGPT. Second, review the settings that govern this setup: Slack plugin access, the connected account, allowed actions, and approvals.

Admins set which apps are available and which actions are allowed in Plugin controls. An app can be available while some actions still need approval. The connected account's own Slack permissions limit what the dot can reach. For example, if the connected account cannot access a private channel, the dot cannot access it either. Approvals can apply to some actions; check plugin permissions to see which actions require approval.

This method is useful when you want the dot to prepare drafts, monitor a channel for you, or gather context without introducing a new participant into Slack conversations.

Give the Dot Its Own Slack Identity

In this setup, your dot appears as a distinct participant in Slack. The owner can message the dot in Slack or mention it in a supported channel. When the dot posts, people in that conversation see the post.

This method requires coordination across three roles. The ChatGPT workspace owner must enable the "Use dots" and "Add dots to Slack and Microsoft Teams" permissions for the intended members, either in the workspace default or through a custom role. The Slack workspace owner or app manager must install the ChatGPT app in Slack and approve app requests if your Slack workspace requires approval for new apps. Finally, each member opens their dot's profile in ChatGPT, selects Add, and connects Slack.

Enabling the ChatGPT permission does not install anything in Slack or connect anyone's dot. It only grants permission. Each member must complete the connection step themselves.

Use this method when you want the dot's contributions to be visible. For example, if your dot posts a weekly project update in a team channel, other members see that the update came from the dot.

Role-Stacking Gotcha

If one role grants dots access and another role does not, turning off the permission in one role will not remove access. To give selected groups access while keeping others out, set the workspace default to off and assign a custom role to the intended members.

Turning Off Slack Access

Turning off a dot's Slack identity does not remove its access through the member's connected Slack account in the plugin layer. When you offboard someone or revoke Slack access, check both the Slack identity setting and the member's plugin authorizations.

Abstract 3D scene of a glossy emerald sphere connected by a cream light cable to a glass box and by a magenta cable to a chat bubble, showing two ways to connect

Who Can Direct the Dot and Who Can See Its Posts

Only the owner can direct their dot through a Slack direct message or by mentioning it in a supported channel. Messages from other people do not start work. If you add your dot to a team channel and another member writes something, the dot does not treat that message as an instruction. However, your dot may use other participants' messages as context when you bring it into a Slack thread.

Anyone with access to the conversation can see the dot's posts. If you add your dot to a public channel, everyone in that channel sees what it writes. If you message your dot in a private channel, only members of that private channel see the conversation.

By default, your dot responds to you in Slack. You can instruct it to engage with others. For example, you might say "When someone asks a question in this channel about the launch timeline, reply with the current schedule and tag me." Existing permissions and safeguards still apply. Your dot can message you privately to ask what you are comfortable sharing before it replies in a channel. OpenAI's guide to how you message your dot in Slack covers direct messages, adding your dot to channels, and forwarding messages for follow-up.

Sharing Boundaries Across Channels

The same dot is reachable through ChatGPT, Slack, and Microsoft Teams. Switching channels does not create a separate dot or reset its saved notes. Your dot maintains context across all the places you talk to it.

Visible conversations stay distinct. A message you send in ChatGPT is not copied into Slack. A Slack thread shows only the messages exchanged in Slack. Your dot can use relevant context from one channel to inform its work in another, but the ability to use information does not grant permission to disclose it to a different audience.

Imagine you share a client deadline with your dot in a private ChatGPT conversation. Later, you ask your dot to prepare a project update for your team channel in Slack. The dot can use the deadline to structure the update, but sharing private details in the team channel still requires permission. Before your dot shares information from a private conversation with others, it checks that you have allowed it.

Write sharing instructions that make the boundary explicit. For example: "You can use the client deadline I shared with you to organize the update, but do not mention the client's name or specific contract terms in the team channel. If you need to reference the client, ask me first."

Another example: "When you post the weekly update in the team channel, include task status and blockers. Do not share budget numbers or client feedback unless I have already posted that information in the channel myself."

Your dot can also message you privately to check what you are comfortable sharing before it replies in a channel. If you want tighter control, add a custom rule that requires approval before posting in certain channels. More on that below.

Proactive Research and Slack Context

Your dot can research information it has permission to read, which can include a connected Slack source, and keep private notes. This proactive research happens in the background. Your dot does not send messages, change connected apps, or control your browser or computer during proactive research. Follow-up actions are subject to your permissions and approval settings.

If your dot has access to a Slack channel, it can read messages in that channel as part of proactive research. It may use that context to prepare better updates or to notice changes that affect your work. Reading a channel does not give your dot permission to post in that channel or to share what it read with others. Posting and sharing still require explicit permission.

Assigned recurring tasks can include actions you have authorized. For example, if you tell your dot to post a weekly update every Friday, that is an assigned recurring task with posting permission. The permissions and schedule for that task are separate from proactive research.

Routing Updates to the Right Channel

You can tell your dot where you want different types of updates. For example: "Keep routine progress notes in ChatGPT. Bring decisions to me in Slack. If a deadline is at risk or you need my input, message me in Slack immediately."

This routing instruction helps your dot understand which changes deserve your attention in Slack and which can stay in the background. A sample instruction: "Check the project plan every weekday morning. Update the task list in ChatGPT when someone completes a task. Message me in Slack only if a deadline slips, a blocker appears, or you need a decision from me."

Routing instructions work best when you specify the trigger and the destination. "If X happens, notify me in Y channel" is clearer than "keep me updated."

Channel Monitoring, Recurring Schedules, and Event Monitoring

Adding your dot to a Slack channel does not establish a monitoring schedule. You must tell your dot which channels to check, what to watch for, and when to notify you.

Channel Monitoring

A monitoring instruction specifies the channel, the condition, and the action. For example: "Monitor the support channel. If someone reports a bug that affects the checkout flow, investigate and prepare a summary for me in ChatGPT. Message me in Slack if the bug is critical."

Ask your dot to confirm the recurring task. "Confirm that you will check the support channel and notify me in Slack if a critical checkout bug is reported." Review the confirmation to make sure the dot understood the scope and timing.

Recurring Schedules

Recurring tasks need a saved schedule. Specify what to check or update, when to run the task including time zone and end date, which changes deserve a notification, and where to deliver the update.

Sample schedule prompt: "Check the planning channel every weekday at 9 AM Pacific for the next four weeks. Update the project checklist in ChatGPT when a deadline changes. Message me in Slack only if a deadline is at risk or you need a decision. Confirm the schedule."

Another example: "Every Friday at 3 PM Eastern, review the week's completed tasks in the project tracker. Prepare a summary and post it in the team channel. Run this schedule for the next eight weeks. Confirm."

Include the time zone explicitly. "9 AM" without a time zone can lead to confusion if team members work across regions.

Include an end date. Open-ended schedules are harder to audit and easier to forget. An end date forces you to review whether the task is still needed.

Event Monitoring

Event monitoring depends on the connected service. Where the connected service supports it, your dot can watch for specific events and respond. For example: "Investigate new bug reports in the connected Slack channel and prepare fixes for review."

Tell your dot which events to watch and when to notify you. Ask it to confirm which events it can follow. The supported event types may vary as features roll out. Connecting Slack alone does not create a monitoring task. You must give the instruction and confirm the dot understood.

A sample event-monitoring prompt: "When a new message in the client-launches channel reports a delay, check whether the staging environment is ready and whether QA has started. If staging moved or QA is delayed, draft a follow-up message for me to review. Confirm which events you can monitor in this channel."

OpenAI's documentation includes an example where a dot in a website-launch channel flagged that staging moved, QA compressed, mobile testing had no owner, and the launch was at risk. The dot said "I drafted a follow-up for your review" and "The draft is ready for you to review. I have not sent it." That example illustrates the draft-first behavior you can request.

You can review recurring tasks and schedules in the Scheduled view in your dot's profile, or ask your dot to list, change, or cancel a schedule.

Your dot can also read and edit your existing ChatGPT Work automations. For more on recurring tasks and event monitoring, see OpenAI's tasks-and-memory documentation.

Writing Effective Monitoring Briefs

A monitoring brief is the instruction you give your dot to watch a Slack channel and notify you when something happens. Effective briefs are specific.

Include the channel name. "Monitor the support channel" is clearer than "monitor Slack."

Include the condition. "If someone reports a bug" is clearer than "if something important happens."

Include the action. "Prepare a summary for me in ChatGPT" tells your dot what to do when the condition is met.

Include the escalation rule. "Message me in Slack if the bug is critical" tells your dot when to interrupt you.

Include the schedule. Use fixed times with time zone and end date, or describe event monitoring where the service supports it.

Sample monitoring brief: "Monitor the client-launches channel. If someone posts a message reporting a delay, read the thread and prepare a summary of the issue in ChatGPT. Message me in Slack immediately if the delay affects this week's launch. Check every weekday at 9 AM, 12 PM, and 4 PM Pacific for the next three weeks. Confirm this monitoring task."

Another example: "Monitor the bug-reports channel. If a new message describes a customer-facing issue, investigate the issue and draft a response for me to review. Message me in Slack as soon as you see it. Confirm which events you can monitor in this channel."

Sample Schedule Prompts

Here are sample prompts for recurring schedules in Slack. Adapt them to your workflow. The posting samples assume you have already approved the format after a draft-only trial.

"Every Monday at 10 AM Pacific, review the planning channel for new tasks added in the past week. Update the master task list in ChatGPT. Message me in Slack if a new task has no owner or no deadline. Run this schedule for the next eight weeks. Confirm."

"Every Friday at 3 PM Eastern, prepare a summary of completed tasks from the project tracker. Include task name, owner, and completion date. Post the summary in the team-updates channel. Run this schedule for the next six weeks. Confirm."

"Every weekday at 9 AM Central, check the support channel for messages describing an escalation. If an escalation was posted in the past 24 hours, read the thread and prepare a brief for me in ChatGPT. Message me in Slack if the escalation is still unresolved. Run this schedule for the next four weeks. Confirm the schedule."

"Every Tuesday and Thursday at 2 PM Pacific, review the design-feedback channel for new comments on the current project. Summarize the feedback in ChatGPT. Message me in Slack if feedback requests a change to the approved design. Run this schedule for the next five weeks. Confirm."

Abstract 3D scene of glowing particles flowing from an emerald sphere through a magenta check-mark gate toward glass message bubbles, one bubble locked with a gold padlock

Draft Permission vs. Post Permission

Asking your dot to prepare something is not the same as giving it permission to post.

"Prepare a weekly project update for me to review" asks for a draft. Your dot will write the update and show it to you in ChatGPT or send it to you in Slack as a private message. It will not post the update in a team channel unless you approve.

"Post the approved update in our team channel every Friday" gives permission to post. If you have already reviewed and approved the format, and you want the dot to post without asking each time, this instruction grants ongoing permission within that scope.

Scope matters. "Post the weekly update every Friday" is clear about who, what, and when. "Post updates when needed" is vague. Your dot may ask for clarification or may guess, and guessing leads to mistakes.

A scoped instruction example: "Every Friday at 4 PM Pacific, post the weekly project update in the team channel. Include completed tasks, upcoming milestones, and blockers. Do not include budget or client names. Run this schedule for the next six weeks."

Another example: "When I send you a message that starts with 'Approved for posting', post the attached draft in the client-launches channel. Do not post anything I have not explicitly approved."

Custom Rules for Slack Posting

Custom rules let you set standing instructions for specific actions. You can configure rules in Settings, Personalization, Custom rules on desktop or web, or in the sidebar under Customize, Custom rules on iOS or Android.

Four options are available: Take action without asking, Take action when you say so, Ask before taking action, and Hand off to you.

A useful rule for Slack: "Ask before taking action for sending messages to customers or external channels." If your Slack workspace includes channels that connect to customers, partners, or vendors, this rule adds a confirmation step before your dot posts in those channels.

Another example: "Ask before taking action for posting in any channel that includes the word 'client' in its name."

Custom rules are instructions your dot tries to follow. They are not hard blocks. Your dot can make mistakes. Do not grant access to sensitive channels or override safety requirements by relying on custom rules alone. Use them as a layer of review, not as a substitute for proper permissions.

Select Open Plugins to review Plugin permissions, which control app actions separately from custom rules.

If your workspace administrator disables custom rules, your saved rules will not apply. Check with your admin if you are unsure whether custom rules are enabled.

Do Not Confuse Your Dot with the ChatGPT App in Slack

The ChatGPT app in Slack and Microsoft Teams is a separate experience. People in supported conversations can ask @ChatGPT to summarize information, draft responses, or use approved tools. It uses configured connections and conversation context and does not inherit a person's file, app, or private-channel access. At launch it does not include memory or use a person's ChatGPT memories. Shared replies are visible to conversation participants.

Your dot is personal. It is owned by one person. It responds to its owner. It uses the context and permissions you have given it, including your connected accounts and your saved notes. When you message your dot in Slack, you are talking to your own agent, not to a shared workspace bot.

If your Slack workspace has both the ChatGPT app and dots enabled, make sure your team understands the difference. Your dot is for your own work.

Reviewing Your Dot's Slack Work

Two views help you track what your dot is doing: Activity and Scheduled.

Activity shows tasks your dot is working on or has completed. Open your dot's profile in the ChatGPT desktop app and select Activity. You will see a list of tasks, including tasks that involved Slack. Select a task to review its progress, files, and results, including background work, and respond to requests for a decision, app connection, sign-in, or approval.

Scheduled shows recurring tasks your dot has saved. This view lists the schedule and the instructions you gave. Check the task's instructions, timing, and destination. Disable or delete recurring tasks you want to end, or ask your dot to list, change, or cancel scheduled tasks.

Reviewing Slack work also means checking that the post actually landed. If your dot says it posted an update in a team channel, open Slack and verify. A completed run does not by itself confirm that the requested result was achieved or delivered, so open Slack and check.

Stopping Your Dot's Slack Work

Pause stops your dot's current main task. It does not stop every delegated task or cancel future scheduled runs. Open a delegated task in Activity to stop it. Open Scheduled to disable or delete a recurring task. Use Resume to continue.

Stopping a task does not undo completed actions. If your dot posted a message in Slack, pausing or stopping the task does not delete or recall the message. You must delete the message in Slack manually if needed.

Deleting your dot does not recall messages already delivered to other people. If you delete your dot, its Slack posts remain in the channels where it posted. Plan for this before you give your dot permission to post in shared channels.

Microsoft Teams and Texting

You can message your dot in Microsoft Teams. According to OpenAI's admin guide, Teams access is currently limited to an invite-only alpha. If your workspace is not part of the alpha, Teams connection may not be available yet.

Texting is listed as coming soon. Calls initiated by your dot are planned for after launch. Check OpenAI's current documentation for availability.

Offboarding Slack Access

When someone leaves your team or you need to revoke Slack access, follow these steps.

First, check dots access in the workspace default and every custom role. If the person has access through multiple roles, turning off one role will not remove access.

Second, review plugin authorizations. Turning off the dot's Slack identity does not remove access through the member's connected Slack account. Open the member's plugin settings in ChatGPT and revoke the Slack plugin authorization.

Third, check active tasks, recurring schedules, and delegated tasks. Open the dot's Activity and Scheduled views and identify any tasks that involve Slack. Decide who will take over those tasks or whether they should be canceled.

Fourth, review website sessions and local computer access if the dot has those permissions. Each access type has its own settings. Revoking Slack access does not automatically revoke other access.

Access changes and deletion are separate actions. Revoking access stops the dot from using Slack going forward. Deleting the dot removes it entirely, but does not recall messages already posted in Slack.

Incident Response

If a dot takes an unexpected action in Slack, follow your incident process and take these steps immediately.

Ask the member to pause their dot. Open the dot's Activity view and review the affected task. Save the task link and available activity records for your incident log.

Check the app or computer permissions the dot has. Did it act within the scope you intended, or did it exceed its instructions? Review the custom rules and plugin permissions you configured.

If the action involved posting in a channel, check whether the post should be edited or deleted. If the post shared information that should not have been shared, follow your data-incident process. Whether an action can be undone depends on the action and the service.

After the incident, review the instructions you gave the dot and the permissions you granted. Tighten the scope or add a custom rule to prevent similar actions.

Compliance and Analytics API Notes for Admins

Compliance API records include user prompts and responses, supported tool-call events, and connected-app calls made through the Model Context Protocol. Dots records use client_id = DOTS. Check which records are available before relying on exports for an audit.

The Analytics API daily usage endpoint identifies dots activity with client_id = DOTS. You can use this identifier to separate dots usage from other ChatGPT usage in your workspace analytics. Tasks a dot starts in Work or Codex count toward those products' usage and appear in their reporting.

Prompt Injection and Slack Channels

Slack channels contain text written by many people. Your dot reads that text when it monitors a channel or when you bring it into a thread. Built-in protections reduce the risk that someone else's message will trick your dot into taking an unintended action, but no system is perfect.

This is one more reason to keep posting behind approval at first. If your dot prepares drafts for your review, you can catch mistakes or unexpected behavior before anything is posted. After you have seen several correct drafts and you trust the monitoring brief, you can grant post permission within a clear scope.

Do not overstate the risk, but do not ignore it. Treat Slack monitoring the same way you treat any other connected app: grant the minimum access needed, review the work, and tighten permissions if you see problems.

Common Mistakes

Adding a dot to a Slack channel without giving it monitoring instructions. The dot will not watch the channel or notify you unless you tell it to.

Assuming other people in a channel can direct someone else's dot. Only the owner can direct their dot. Other members' messages do not start work.

Giving vague posting instructions. "Post updates when needed" leaves too much to interpretation. Specify who, what, when, and where.

Forgetting to check both the Slack identity and the plugin authorization when offboarding. Turning off one does not turn off the other.

Not reviewing the dot's work in Slack. Activity and Scheduled views show tasks your dot has saved and is working on, but you should verify that posts actually landed in the right channels.

Granting post permission without a trial period of draft-only. Start with "prepare for review" and move to "post without asking" only after you have seen several correct drafts.

Using custom rules as a substitute for proper permissions. Custom rules are instructions, not enforcement. Do not grant access to sensitive channels and rely on a rule to block mistakes.

Not specifying time zones in recurring schedules. "9 AM" means different things to people in different regions.

Not setting an end date on recurring tasks. Open-ended schedules are harder to audit and easier to forget.

Channel Pilot Checklist

Run a pilot in one Slack channel before rolling out dots across your workspace. Here is a checklist.

  • Choose a low-risk channel. Pick a channel where mistakes will not reach customers or expose sensitive information. An internal project channel or a team coordination channel works well.
  • Decide which connection method to use. If you want the dot's posts to be visible, give it a Slack identity. If you want the dot to work through your own Slack account, connect your Slack account through plugins.
  • Complete the setup chain. Workspace owner enables permissions, Slack owner installs the ChatGPT app if needed, each member connects their dot from its profile.
  • Write a monitoring brief. Specify which channel, what to watch, when to notify, and where to deliver updates. Ask the dot to confirm.
  • Start with draft-only permission. Instruct the dot to prepare updates for your review, not to post them. Review several drafts before granting post permission.
  • Set a time zone and end date for recurring tasks. Make the schedule explicit and time-bound.
  • Add a custom rule for the pilot channel. For example, "Ask before taking action for posting in the project-alpha channel."
  • Review Activity and Scheduled daily during the pilot. Check that tasks are running as expected and that posts land in the right place.
  • Verify posts in Slack. Do not rely on Activity alone. Open Slack and confirm the post appeared.
  • Collect feedback from other channel members. Ask whether the dot's posts are helpful, whether the timing is right, and whether anything should change.
  • Document what worked and what did not. Write down the monitoring brief, the schedule, the custom rules, and any mistakes. Use this documentation to refine your approach before expanding to other channels.
  • Run the pilot for at least two weeks. One week may not surface edge cases. Two weeks gives you enough cycles to see how the dot handles routine updates and unexpected events.

When to Use Dots in Slack and When to Use a Different Tool

Dots are personal agents. They work for one owner. They are not customer-facing bots.

Use your dot in Slack when you need it to monitor internal channels, prepare updates, route decisions to you, or post recurring summaries in team channels. Use it to reduce the manual work of checking channels, copying information, and writing status updates.

Do not use your dot as a customer-facing agent in Slack. Customer-facing conversations belong in GoHighLevel, where Conversation AI handles customer chats across SMS, email, web chat, WhatsApp, Facebook, and Instagram, and Managed Agents automate CRM-triggered workflows. Conversation AI is built for customer conversations and should be designed with its own escalation rules.

Your dot can help you prepare responses to customers, but it should not be the front line of customer communication. Keep customer-facing work in systems designed for that purpose.

SumoGrowth Implementation Help

Setting up ChatGPT dots in Slack involves decisions about which connection method to use, which permissions to grant, how to write monitoring briefs and sharing rules, and how to run a pilot without leaking information or losing control. SumoGrowth Labs helps teams plan the setup, configure permissions, write the briefs, and run the pilot.

We are an AI implementation and consulting agency. We do not sell a product. We help you use the tools you already have or the tools you choose. If you want help planning which Slack setup to use, writing monitoring briefs and sharing rules, configuring permissions, or running a channel pilot, reach out. We will walk through your workflow, identify where a dot can reduce manual work, write the instructions, and help you verify the setup before you roll it out to your team.

by Darius Gaynor

Share this articleLinkedInX

Turn your next idea into growth.

Book a discovery call