Customer conversations (two-way chat)

AI-assisted customer chat, escalation to human, conversation status

Customer conversations

A customer conversation is a two-way chat that starts from a booking email or the online booking page. AI tries to answer first, you can escalate to human reply if needed.

Where to find it

/manager/inboxConversations tab.

How it starts

The customer:

  1. Receives a booking email with a "Have a question? Chat with us" button
  2. Clicks
  3. Token-based chat link opens
  4. Writes a message

The new conversation appears in the Conversations tab with "Active" status.

Conversation statuses

Active

Conversation open, messages flowing.

Escalated

You (or another staff) took over because AI couldn't solve it. AI no longer writes, only you do.

Closed

End of conversation. Neither AI nor you write. But: if customer writes again, auto-switches to Active.

AI auto-replies

Bookinda AI (OpenAI-based) tries to answer:

  • "When is my booking?" → AI checks Bookinda data, answers: "Your booking is 2026-05-05 14:00."
  • "Can I move it?" → AI helps or suggests the customer portal link
  • "How much is a haircut?" → AI quotes from the salon price list

When AI CAN'T reply

  • Complex complaint ("didn't like the service")
  • Payment dispute
  • Custom request (e.g., "can I bring a dog")
  • Apology or compensation needed

→ You escalate and respond as a human.

Message thread

Chronological chat, each message labeled:

  • Customer
  • AI ✨
  • Staff 👤

Escalation

"Escalate" button on the conversation header. After clicking:

  • AI no longer replies
  • You get push / inbox flag
  • Status: Escalated

Closing

"Close conversation" button. Goes to Closed. If customer writes again, auto-switches back to Active and you get notified.

Token expiry

A conversation token lives 30 days. If customer writes from the old link after 30 days, a new conversation starts.

When to use it

Scenario 1, "When is my booking" AI auto-answers, you get a heads-up: "AI replied, check if you want." Usually no action needed.

Scenario 2, "I want to cancel" AI helps: "Here's the customer portal link to cancel, or we can do it. Confirm?" If customer confirms, you cancel.

Scenario 3, "Didn't like the service" AI may auto-reply: "Sorry to hear. A colleague will reach out shortly." You escalate and respond personally.

Scenario 4, payment dispute AI won't touch this. Mandatory escalation. Review customer transactions and respond personally.

Scenario 5, customer writes again on a closed conversation Reactivates. You see the old thread and continue from there.

Access

Only visible inside the salon. Per-role:

  • Owner / Manager / Reception: every conversation
  • Staff: only conversations linked to their bookings

Tips

  • Read AI replies periodically, to see if it works well. First month tuning matters.
  • Escalate fast if AI misreads. Customer doesn't like wrong answers.
  • Don't reopen Closed conversations yourself, only when customer writes. Otherwise the "closed" feeling vanishes.
  • 30-day token is long enough, but if customer writes later, new conversation. Not necessarily bad.
  • Don't use customer conversations for marketing, it's two-way help, not promotion.
  • Keep push notifications on, so you don't miss an escalated conversation.

Related articles

#conversations#chat#beszélgetés#ai#escalate
💬

Have a question? Ask the AI assistant

The Bookinda AI knows the entire help center and can answer in seconds.

B

Bookinda AI Assistant

Ask anything about features, settings, or integrations.

Hi! I'm the Bookinda assistant. I can answer questions about features, billing, integrations, and more. What would you like to know?

Still need help?

Get in touch with our support team.

Contact us