Work the conversation where the team works. Answers from your docs. Team replies in Slack or Telegram.

Support gets missed when it lives somewhere else

A second helpdesk can work well when a company has a dedicated support operation. In a small SaaS team, it can also become another tab with its own queue, assignment rules, and notification settings. Meanwhile, engineers discuss incidents in Slack and founders answer customer questions wherever they see them first. The risk is not Slack itself; it is unclear ownership when a visitor needs a real person.

Visitors may ask “Can someone review my failed import?”, “Why does my workspace show 80% complete?”, “Can you enable SSO for our account?”, and “Where is the API limit documented?” A public docs answer can resolve the last question. The first three may require private context, so they should reach the people who can investigate rather than receive a generic guess.

• Create a clear team channel and owner for handoffs.
• Use the public documentation for repeatable product facts.
• Keep sensitive account work out of a public answer.

One visitor conversation becomes a Slack thread

When a question needs a person, MivoChat sends the conversation to Slack as a thread. The messages stay grouped with their customer context, and a teammate can reply without asking the visitor to copy the problem into a different system. The reply appears live in the website widget. Your team can take over, request more detail, hand the exchange back to AI, or close it when the question is resolved.

MivoChat can also answer first when the connected docs clearly support the response. Connect the plan page, onboarding instructions, and troubleshooting guides. If a page does not establish an answer, the assistant should say it does not know and hand off. Set channel notifications and ownership expectations before launch so threads do not accumulate without a person responsible for them.

Keep the economics understandable

Starter is $29/month with 3 seats, one site, and 1,000 AI replies. Pro is $99/month with 10 seats, five sites, and 10,000 replies. The limit counts AI messages, not conversations. The default support estimate uses 3 replies per AI-handled conversation, so the Starter allowance covers about 333 conversations at that average.

Fin at $0.99 per resolved conversation would cost around $329.67 for 333 conversations, before seat fees. MivoChat has no per-resolution fee; its cap is measured in replies, and a conversation that needs more than three replies uses more of the allowance. Once the plan limit is reached, new questions go to a human without another model call. Prices are published, but paid checkout is still coming soon. A 14-day trial is available without a card.

Install the website side before connecting Slack

Add the widget to your public site and docs pages using the provided workspace site key. For a standard site, the snippet is: <script src="https://mivochat.app/widget.js" data-site="YOUR_SITE_KEY" async></script>. For Next.js, add the script in your layout with next/script. The visitor-facing widget is still part of your website; Slack is the place your team handles human replies.

After installing, connect Slack and send a test handoff. Verify that the expected channel receives a new thread with the right conversation, that a teammate reply reaches the widget, and that take-over, return-to-AI, and close behavior are available. Test a documented product question as well, and make sure it can be answered without sending every conversation to the team.

Slack is the handoff channel, not a private data connector

MivoChat does not inspect a customer account, change permissions, enable SSO, investigate an import, or perform any product action. Keep the answer language precise and do not imply that Slack access gives MivoChat access to your SaaS data.

Your team remains responsible for account-specific billing, incidents, and access requests. If docs are insufficient, the assistant should say so. The handoff route only works if channel owners monitor it, so pick a practical alert setup and define who takes over when the usual owner is away. Close conversations when appropriate to keep the working queue useful.