Using Agents in Slack

Doti Agents are now fully integrated into Slack, bringing domain-specific reasoning and customized workflows directly into the flow of conversation. With agent support, users can choose the right assistant for the job - from Customer Prep to IT helpdesk - without ever leaving Slack.

This guide explains how to select, switch, and use agents seamlessly within Slack.


Choosing an Agent in Slack

There are two ways to use an agent when chatting with Doti in Slack:

Before You Ask

  • Use the agent selector in the message composer to pick which agent you want to use.

  • Your selected agent will handle your next question - with all its custom logic, tone, and data access.

After You Ask

  • If Doti replies using the wrong agent, you can change the agent post-response.

  • The conversation will rerun using the newly selected agent, instantly updating the reply.

🧠 Doti remembers your agent selection per user. It will continue using that agent until you choose a new one.


How Agent Identity Appears in Slack

When you use an agent in Slack:

  • The agent name is shown as the responder (e.g. “Customer Prep Agent”)

  • The agent emoji is also updated, making the persona instantly recognizable

  • In Autopilot mode, Slack replies will show the correct agent identity too

This creates clarity in shared channels - helping teams know who’s “speaking” and what kind of logic is applied.


Real Use Cases in Slack

Agent Name
Slack Purpose

🧯 On-Call Agent

Responds to incident summaries or root cause questions

💼 GTM Agent

Helps Sales & Marketing pull messaging, competitive intel, and positioning

🧠 Default Agent

Handles general questions and workflows across the org

🛠 IT Support Agent

Answers tooling, permissions, and internal process queries

🤝 Customer Prep Agent

Builds full customer briefs ahead of meetings


Autopilot + Agents

When Autopilot is enabled in a channel:

  • Doti will respond as the agent assigned to that Autopilot

  • The Slack identity (name + emoji) will reflect the chosen agent

  • All replies follow that agent’s instructions and data scope

✨ This makes Autopilot smarter, clearer, and more context-aware across large team channels.


Behind the Scenes: Agent Instructions Still Apply

When you interact with an agent in Slack, it uses all of its backend configuration:

  • Custom instructions (reasoning steps, tone, formatting)

  • Assigned collections and permissions

  • Personality and context logic

  • Tool usage (e.g. web search, knowledge filters, document parsing)

You get the full agent power, right inside Slack.


Summary

  • Agents can be selected before or after asking a question

  • Agent identity appears clearly in Slack with name + emoji

  • Autopilot also supports agent identities in responses

  • Agents behave the same in Slack as they do in the Portal or Chrome Extension

  • Behind the scenes, they apply full instruction logic and permissions


Need help configuring agents for Slack use? Visit our Agents section to learn more.

Last updated

Was this helpful?