Skip to content

Agents

Add a conversation agent

Create a conversation agent that chats with users, grounded in your workspace's documents and resource libraries.


A conversation agent is the platform’s core agent type: it chats with users and answers their questions. You can ground its answers in your workspace’s knowledge — the documents you’ve uploaded — and let it surface entries from your resource libraries during the conversation.

Agents live in your workspace and are created from the New Agent card on the workspace overview (or the + next to Agents in the left sidebar). Conversation is the default type.

How the editor works

Once created, the agent opens in an editor with tabs — for a conversation agent: General, Model, Sources, Resource libraries, and (when they’re available) Conversation categories, Orchestration and Embed. Each tab is its own form and saves on its own — click Save on a tab to keep its changes; the editor warns you before you leave a tab with unsaved changes.

The full flow at a glance

The walkthrough below replays every step in the real interface. Use Prev / Next to move at your own pace; each step highlights the button to click and the area to watch.

DemoStudio
Conversation Bayes Assistant
Extraction Extraction Agent
New Agent

An agent can perform specific tasks based on its type and configuration.

Create
Edit Conversation AgentEdit details for this conversation agent
GeneralModelSourcesResource librariesConversation categoriesOrchestrationEmbed

Name

Helpful Assistant

Locale

EN

Greeting

Hi! How can I help you today?

Instructions

You are a helpful assistant. Answer the user's questions clearly and, when relevant, base your answers on the workspace's documents.
Save
Edit Conversation AgentEdit details for this conversation agent
GeneralModelSourcesResource librariesConversation categoriesOrchestrationEmbed

Model

Select model

Temperature

0.7
Save
Edit Conversation AgentEdit details for this conversation agent
GeneralModelSourcesResource librariesConversation categoriesOrchestrationEmbed

Answer source

Search by document tags

Document Tags

Getting started Product
Save
Edit Conversation AgentEdit details for this conversation agent
GeneralModelSourcesResource librariesConversation categoriesOrchestrationEmbed

Resource libraries

Manage libraries

The assistant can surface resources from the selected libraries when a user's request matches a resource's title or description.

Getting started Add library
Save
Edit Conversation AgentEdit details for this conversation agent
GeneralModelSourcesResource librariesConversation categoriesOrchestrationEmbed
Product Pricing Support
Select all

Categories already used in this agent's conversations cannot be unchecked.

Save
Edit Conversation AgentEdit details for this conversation agent
GeneralModelSourcesResource librariesConversation categoriesOrchestrationEmbed

Sub-agents

Extraction AgentEnabled

Tool name

ask_extraction_agent

Tool description

Route document-extraction requests to Extraction Agent.

Available conversation agents

Conversation Agent Add
Save
Edit Conversation AgentEdit details for this conversation agent
GeneralModelSourcesResource librariesConversation categoriesOrchestrationEmbed
Enable embed

Embed snippet

<script src="https://app.example.com/launcher.js" data-token="agt_1a2b3c"></script>

Allowed origins

https://app.example.com

Widget title

Helpful Assistant

Logo URL

https://…/logo.png

Brand color

#2563eb
Update

Step by step

1. Open the New Agent dialog

On the workspace overview, find the New Agent card and click Create (you can also use the + next to Agents in the left sidebar).

2. Keep the Conversation type

In the New Agent dialog, under Agent type, keep Conversation (selected by default).

3. Name the agent and create it

Type a Name (at least 3 characters), then click Create. Your agent is created with sensible defaults and opens in the editor.

4. General

Set the Name, the Locale (the language the agent works in), the Greeting (the first message it sends) and the Instructions (what defines its behaviour and tone). Click Save.

5. Model

Open the Model tab, choose the Model and a Temperature (higher = more varied, lower = more focused). Save.

6. Sources

Open the Sources tab and set the Answer source:

  • Answer from model knowledge only — no documents consulted.
  • Search across all documents — every document in the workspace.
  • Search by document tags — only documents with the Document Tags you pick.

Save. See Documents to upload and tag documents first.

7. Resource libraries

Open the Resource libraries tab to attach the libraries the agent can surface during a conversation. Click Add library, pick one, and it appears as a tag. Use Manage libraries to open the library manager. Save. See Resource libraries to create them first.

8. Conversation categories

If your workspace has conversation categories, open the Conversation categories tab and tick those that apply to this agent (Select all adds them all). Categories already used in this agent’s conversations can’t be unchecked. Save.

9. Orchestration

When orchestration is enabled, open the Orchestration tab to add sub-agents this agent can delegate to. Pick from Available conversation agents, give each a Tool name and Tool description, and enable it. Save.

10. Embed

When embedding is enabled, open the Embed tab to put the agent on an external site: toggle Enable embed, copy the Embed snippet, and set the Allowed origins, Widget title, Logo URL and Brand color. Click Update.

Tips

  • Write the Instructions in the same language as the agent’s Locale.
  • Use Search by document tags to keep answers scoped to the right documents in a large workspace.
  • Remember to Save each tab — changes on one tab aren’t kept when you move to another without saving.
  • Conversation categories, Orchestration and Embed only appear when the workspace has categories or those features are enabled — you may see fewer tabs.

Troubleshooting

  • The Create button is disabled — the Name must be at least 3 characters.
  • The agent ignores my documents — check the Answer source on Sources; “Answer from model knowledge only” consults no documents.
  • No tags to choose — the Search by document tags option and the Document Tags picker only show once your documents are tagged; see Documents.
  • My changes disappeared — each tab saves separately. Click Save (or Update on Embed) before switching; the editor warns you when changes are unsaved.
  • I don’t see the Orchestration or Embed tab — they’re optional features; ask your workspace Owner or an Admin if they aren’t enabled.

Last updated: July 13, 2026