AI agents streamline your internal processes by delivering instant, accurate answers drawn from your company’s existing knowledge base and connected resources. By automatically handling routine employee inquiries, it reduces the need for human intervention, saves valuable time, and ensures team members receive timely assistance 24/7, even outside business hours. This enables your staff to focus on higher-value tasks while maintaining a more efficient and responsive workplace.
Note: Every time you update information in the knowledge base, it will automatically update the AI agent’s knowledge.
How To Create an AI Agent
To create an AI agent, follow these steps:
Navigate to the General Settings by clicking on your avatar on the top right.
Select Settings from the menu.
Enter the AI Agents tab and click Create agent
Select the knowledge bases the AI agent will have access to, which it will use to base its answers. Connect to the Knowledge Base that you have already created in Connecteam.
Note: Links to public websites will also be included in the agent’s knowledge. URLs like Google Drive and YouTube will not work.Next, name your AI agent and describe what it's here to help with.
Optionally, add real-life example questions in the Customize agent step to show employees what they can ask the agent and help guide more accurate, relevant responses.
Select the users and groups who will have access to the AI agent
Click Create
If you want to add additional agents after creating your first one, click Create agent in the top right and follow the same steps outlined above.
Usage Dashboard
The Usage Dashboard gives admins full visibility into how employees interact with AI Agents.
In the AI Agents settings, open the new “Usage” tab to view:
Total conversations
Total users
Total messages
You’ll also see a detailed table where you can:
Search usage data
Filter by agent
Filter by date range
Setting Up Custom Instructions
In the Customize agent step, admins can now add Custom Instructions to control how an AI Agent responds without updating the Knowledge Base.
You can set up (or update) Custom Instructions in either of these ways:
During the setup wizard when creating a new AI Agent (in the Customize agent step).
After the agent is created, by going to General Settings → AI Agents, clicking the three dots (⋯) next to the agent, and selecting Customize.
Use Custom Instructions to set rules like:
What the agent should focus on (e.g., “Answer only IT help topics like password resets, Wi-Fi, and VPN,” or “Focus on onboarding questions like first-day steps and required training”).
What the agent should avoid (e.g., “Don’t suggest disabling security settings,” or “Don’t answer anything related to payroll or legal advice—direct employees to HR instead”).
How the agent should behave (tone and guardrails) (e.g., “Use a calm, step-by-step troubleshooting style,” or “Use a friendly onboarding tone and ask one clarifying question if the employee’s role isn’t clear”).
Editing your AI Agent
To edit an AI Agent, go to General Settings (click your avatar in the top right) and open the AI Agents tab. Find the agent you want to update, click the three dots (⋯), and select Edit.
In the pop-up window:
Choose the Knowledge Bases the AI Agent can access.
Update the agent’s name and description.
In the Customize agent step, you can:
Add Custom Instructions to guide how the agent responds (tone, focus areas, and guardrails).
Add FAQ example questions to help employees understand what they can ask the agent.
Select the users and groups who should have access to the AI Agent.
When you’re done, review the details on the summary page. If everything looks correct, click Save Changes.
How It Will Appear on Your Employees' Mobile Devices
Once your employees enter their mobile app and click on the chat icon at the bottom of their screen, they will see the AI agent chat, marked with an orange 'sparkle' icon. When they enter the chat, they can ask their questions directly.
💡Tip: by clicking on view sources, users can access the knowledge base from which the AI took the information.
Supported vs. Unsupported Content in the Knowledge Base
Below you can find a list of content types that the AI agent supports when pulling information from your Knowledge Base, and those that the agent does not support.
Supported Content | Content That is Not Supported |
Text objects: Any knowledge base text object is fully supported. | Any link that requires authentication (SSO/login, cookie/session-gated, IP allowlist, or VPN). |
Files (uploaded to the knowledge base): Supported file formats include: xlsx, xls, pdf, doc, docx, csv, jsonl, txt, md, html, json, xml, yaml, yml, jpg, jpeg, png, gif, webp, pptx, ppt | External file URLs (e.g., direct links to .pdf, .docx, etc.). Upload files into the knowledge base instead. |
Links (public only): Public web pages that start with http:// or https:// and do not require authentication, and are not from restricted hosts (see below).
Examples of supported links (must be public/no login) - Public company/Help Center pages - Public vendor docs - Public sections of vendor KBs/portals (when accessible without login) | Links from blacklisted domains, including: - Search engines (Google, Bing, Yahoo, etc.) - Video/Media streaming (YouTube, TikTok, Instagram, etc.) - Social media (Facebook, Twitter/X, LinkedIn, Reddit, etc.) - Temporary file-sharing (WeTransfer, SendGB, etc.) - Cloud storage with expiring/restricted links (Dropbox, OneDrive, iCloud, Mega, etc.) - News sites with paywalls/restrictions (NYTimes, WSJ, Bloomberg, etc.) Other non-relevant sites like e-commerce, entertainment, adult content, dating apps, etc. |
| Files without a valid extension or outside the whitelist. |
Deactivating your AI Agent
If you need to turn off your AI agent, follow these instructions.
Click on the three dots next to the AI agent of your choice.
Select Deactivate.
The status of the AI agent will change to Inactive.
To reactivate the AI agent, click on the 3 dots again, and select Activate.
The status will change back to Active
Frequently Asked Questions
Is any information from our Knowledge base shared externally when the AI generates a response? How does Connecteam ensure that company data remains secure and is not transferred to third-party platforms during AI processing?
Is any information from our Knowledge base shared externally when the AI generates a response? How does Connecteam ensure that company data remains secure and is not transferred to third-party platforms during AI processing?
Only minimal relevant data is transferred to our LLM providers (Azure OpenAI/Gemini/OpenAI) as needed to process the AI requests. Under our strict privacy and security agreements with these providers, we operate under zero data retention policies - meaning no data is stored, saved, or retained on those LLM platforms after processing. Additionally, the data is not used for training or improving their models. All processing is conducted under secure conditions in compliance with stringent privacy and data security standards.
* AI Agents are available from the Expert plan on the Communications hub, with the ability to create up to 20 agents, and is not available for the SBP plan *
Related Articles:
Need more guidance? 🙋 Our LIVE support team (at the bottom right corner of your screen) replies to ANY question.