Skip to main content
Use YCloud Inbox to centralize WhatsApp inquiries, share context, assign work, and continue conversations across handoffs. AI Agents can answer using configured knowledge; rule-based agents can handle fixed questions and flows. Both need explicit human takeover arrangements. Use this for product questions, orders, appointment changes, and after-sales support. Start with one entry point, routing setup, and complete transfer before adding more automation.

​
1. Invite the team and prepare reception

Connect a number, then invite members through Settings → Users and teams → Invite user and assign suitable roles and teams. Members activate their invitations, sign in to Inbox, configure notifications, and set themselves to Available when ready to receive work.
Invite user dialog highlighting email and role, with optional team and capacity settings below.

Choose the invited user’s role and team before sending the invitation.

An invitation does not add the member to the number’s assignment rules. Include the team or member there and check permission to initiate conversations from each number. See the Admin guide and Customer support guide.

​
2. Configure assignment and waiting messages

Open WhatsApp accounts → Settings → Assignment, confirm the number, and decide:
  • Whether the customer owner or previous agent has priority.
  • Which team, person, or Agent receives new customers, and where unmatched conversations go.
  • How to handle unavailable teams and non-working hours.
If your account supports advanced rules and you need additional conditions, follow Conversation assignment to check, save, and enable them. Under Settings → Automation, configure Ice breakers, unassigned replies, queue replies, and delayed-response notices as needed. Set the waiting times and content, then save. Avoid overlapping opening replies. See Automation.

​
3. Add automated reception when useful

AI Agent: questions that need knowledge and context
  1. Open AI Agent → Create Agent and choose an appropriate role, such as Customer service.
  2. Configure business context, response instructions, knowledge, and required Skills. Specify escalation cases and information the Agent must not invent.
  3. Configure Actions such as escalation. Test common, unknown, and transfer questions in Preview.
  4. Select Publish, then assign the published Agent under Settings → Assignment → Assign to AI Agent for the number.
Profile and Skills changes require publishing. Knowledge changes take effect immediately, so consider their effect on live answers. See Create and configure an AI Agent. Rule-based Agent: fixed steps Create a Rule-based Agent and define fallback behavior. Under Create a flow, add keywords with exact or containing matching, connect questions, buttons, and preset messages, and configure human transfer. Enable the configured flow through the current interface and confirm it actually handles the target number. See Create a flow.

​
4. Create a customer entry point

Under the number’s Settings → Chat Links:
  1. For links or QR codes, enter an internal name and prefilled message, then copy the link or download the QR code for your app, website, social profile, or printed materials.
  2. For a widget, select the number, configure message and appearance, and install the generated code on your website.
  3. Enable Source tracking when agents need to identify the originating product page. It captures the URL at widget opening, not browsing history.
  4. Open the entry point as a customer and send a message. Check the correct number and Inbox. If editing regenerates a link or code, replace the installed version as instructed.
Customers still need to send prefilled text. Displayed greetings at an entry point are separate from outgoing WhatsApp messages. See Chat Links.

​
5. Respond, transfer, and close

Open the assigned conversation, inspect history, customer details, and source, then reply. Add useful tags or update contact attributes. Prepare common answers using the Canned responses setup. Use Transfer at the top of the conversation for human takeover or reassignment, choose the recipient, and confirm context is visible. Merely sending a human reply may not stop AI replies; complete the explicit transfer. Select Close after handling the issue. Configure and test auto-close or CSAT if used; current CSAT automation applies to closed human-handled conversations. Use an appropriate template for permitted proactive follow-up after the service window closes. See Service messages.

​
6. Test and improve

Have a test customer use the entry point and check assignment, queue messages, Agent responses, human transfer, closure, and follow-up. With coexistence, also check app-reply synchronization and how the team prevents duplicate responses. Use Inbox → Analytics to inspect unanswered and unassigned work, agent status, first-response and handling times. Review Conversation logs for transfer quality and actual resolution. Use AI Performance to refine automated knowledge and escalation. See Inbox Analytics for metric definitions. Evaluate timely responses, resolved issues, and customer feedback, rather than closures alone.