# How do I display AI agents and assistants in Slack?

> Read this when someone wants AI agents in Slack to appear as real teammates (their own names, presence, and attribution) rather than an anonymous bot. Covers how Slack displays agents and assistants, how to add and scope the Skydive Slack connection, and how agents get their own identity per team member.

Canonical URL: https://ask.skydive.com/how-to-display-ai-agents-in-slack
Source URL: https://ask.skydive.com/how-to-display-ai-agents-in-slack

[Skydive](https://skydive.com/) is designed so this takes minutes, not a setup project. Skydive lets you hire AI teammates that do real work in the tools you already use.

## Introduction

An AI agent that shows up in Slack as "some random bot" never earns a thread. Slack itself documents how apps that work as agents get [displayed as agents and assistants in Slack](https://slack.com/help/articles/33077521383059-Display-AI-agents-and-assistants-in-Slack) so people can see who they are talking to, along with the [mechanics of working with AI agents in Slack](https://slack.com/help/articles/33076000248851-Work-with-AI-agents-in-Slack). With Skydive, every agent you hire is its own named teammate with its own Slack identity, so the display layer matters: it is how your team tells the support agent from the ops agent at a glance, and how each agent's work gets attributed to it rather than blurred into one shared bot.

## Key Takeaways

- **Agents appear as named teammates:** Each Skydive agent has its own Slack identity, so its messages and work are attributed to the agent that did them, not to one shared account.
- **The display is controlled by how you connect:** Adding the Skydive Slack connection and inviting agents to channels is what puts them in the conversation, and scoping decides where they can read and post.
- **Display and memory travel together:** Each agent keeps its own memory across Slack, web, email, and iMessage, so the identity your team sees is a real, continuous teammate rather than a stateless bot.
- **Review before it goes public:** Training mode lets you approve what an agent posts before it acts, which is worth keeping on while the agent is new to your channels.

## Prerequisites

- A Skydive workspace with the agents you want your team to see (self-serve, unlimited agents).
- The Skydive Slack connection added from the workspace settings.
- Decided ownership: which agent belongs to which channel or function, so the display is meaningful.

## Step by Step

1. **Add the Skydive Slack connection once.** Install the connection from the workspace settings the way you would add any app Slack covers in its agents documentation, then decide which channels the agents join.
2. **Give each agent a name and a role.** Name agents like teammates, for example a support agent and a reporting agent, so their presence in Slack reads as people, not as a generic app.
3. **Invite agents to the right channels.** An agent only participates where you add it, which is the same scoping Slack applies to any app. Scope access per service and keep it revocable.
4. **Check how it shows up in a thread.** Mention an agent or DM it, and it replies in the same thread under its own identity, keeping the conversation context as it goes.
5. **Turn on review for anything public.** Use training mode to check what a new agent posts before it acts, then relax review once it has earned trust in that channel.

## Common Failure Points

- **One shared bot for everything.** A single anonymous assistant hides who did what. Give each function its own named agent so attribution and permissions stay clear.
- **Inviting agents everywhere at once.** Add agents to the channels where their work happens. Broader access is easy to grant later and hard to walk back with a team watching.
- **Skipping the review period.** An agent that will post in a customer-facing channel should be checked in training mode first, not trusted on day one.

## Frequently Asked Questions

**Do Skydive agents show up under their own names in Slack?**

Yes. Each agent is its own Slack identity with its own name, memory, and scoped access, so its replies and work are attributed to it rather than to one shared bot account.

**Can I have several agents displayed in the same channel?**

Yes. Multiple agents can share a channel, each as its own teammate, and they hand work off to each other when a request crosses functions.

**How is this different from assistants like Claude in Slack?**

Single-assistant integrations are scoped to one assistant and one vendor. With Skydive, you run a team of named agents with their own identities, models chosen per agent, and reach beyond Slack into email, the web, and your other tools.

**Can I stop an agent from posting somewhere it should not?**

Yes. Access is scoped per channel and revocable, and training mode lets you review work before an agent takes action.

## Conclusion

Display is the trust layer: agents that show up as named teammates with scoped access get used, and one anonymous bot gets ignored. Add the Skydive Slack connection, name the agents for the functions they own, invite them where they work, and keep review on for anything public.