Companion is now on Google Play!GET IT ON GOOGLE PLAY
İçerik Yöneticisi logo

İçerik Yöneticisi

Digital Product Studio

Example scenarioCustomer Communication

Smart Customer Support and Request Handling

A support layer that answers questions from your website, forms and messaging channels using only information you have approved — and passes anything it cannot answer to the right team with a short summary.

Smart Customer Support and Request Handling

The problem

Questions arrive on every channel, and answers depend on who replies

Much of a support team's day goes on answering the same questions again: when will my order ship, how do returns work, what does this plan include. Requests that arrive out of hours wait until morning, and a complex one can land with the wrong person and change hands several times. Customers wait, the team gets worn down, and the same question can get two different answers.

  • Repeat questions eat up the team's time
  • Out-of-hours requests go unanswered
  • Requests don't reach the right person the first time
  • Different people give different answers to the same question

The solution

Known answers straight away, everything else to the right person

The assistant works only from sources you approve: FAQs, returns and delivery policies, product and plan details. When the answer is in those sources, it replies. When it isn't — or the request is sensitive or account-specific — it doesn't guess. It summarises the situation and routes the request to the relevant team by topic, so they can pick it up without making the customer start over.

Possible benefits

  • Consistent answers to common questions, at any hour
  • Your team's time goes to the requests that genuinely need a person
  • Requests reach the right team first time, with a topic and summary
  • You can see which questions come up most and fill gaps in your information

How it works

  1. 01Customer

    A customer asks a question

    Through the website chat, the contact form or a connected messaging channel.

  2. 02AI

    The assistant checks approved sources

    Answers draw only on the policies, FAQs and product information you provide — nothing outside them.

  3. 03AI

    It answers, or classifies and summarises

    If the answer exists, it replies. If not, it tags the request with a topic, urgency and short summary.

  4. 04System

    The request reaches the right team

    It is passed to your help desk or inbox according to rules you set.

  5. 05Your team

    Your team takes over and resolves it

    Refund approvals, complaints and account-specific decisions always stay with people.

Example: an online cosmetics shop

One evening a customer asks which day their order will be dispatched and whether they can return a product they have opened. The assistant answers the delivery question from the shop's delivery page. Returns of opened items depend on exceptions, so it makes no promise; instead it passes the request to customer service tagged "return – needs review", along with a summary of the order. The next morning the team resolves it without having to ask the customer anything again.

What it needs

Information required

  • Up-to-date FAQs and their answers
  • Returns, exchange, delivery and privacy policies
  • Product or plan information
  • Request types and the rules for which team handles each

Systems to connect

  • Website chat window and contact form
  • Your existing help desk or ticketing system (connected during setup)
  • Messaging channels such as the WhatsApp Business Platform (optional, connected during setup)

Where a person approves

Refund and compensation decisions, complaints, legal or health-related questions and account-specific actions are never settled by the assistant; they are summarised and handed to your team. Your team decides which topics the assistant answers and which it stays out of.

Security and privacy

  • AI calls are made server-side; the API key never reaches the browser
  • The assistant uses only the sources you define; any access to customer data is set up separately, with the least privilege needed
  • Usage limits, abuse protection and a one-step off switch are part of the setup
  • How long conversation logs are kept follows your own personal data policy

What it does not do

  • It cannot know what isn't in its sources; incomplete information means incomplete answers
  • AI is not error-free and can occasionally misunderstand, which is why critical topics are left to people
  • It does not replace your support team; it takes the repetitive load off them
  • It does not take payments or cancel orders on its own

How much of your support load is the same questions?

Let's start from real examples of recent questions and work out together which ones an assistant can answer safely and which should stay with your team.

This is an example scenario, not a completed client project. Setup and results depend on your business data and the systems you use.