Meta Pixel

Free framework · Department

The best automations to set up in customer service

The customer service jobs to move to Claude or ChatGPT first, starting with your voice and your policies, so every draft sounds like your team at its best.

For customer service leads and small support teams who answer customers by email, chat and social from a helpdesk such as Zendesk, Gorgias or Freshdesk, often with Shopify or an order system behind it. These are the jobs we would move to Claude or ChatGPT first.

The 7 automations, in the order we would start them

Each one recurs every week or month, reads a system you already run, and has a person who already checks the result. That is why we start here: nothing goes out or gets posted on its own, and the check that keeps the work honest stays with your team.

  1. 01

    Brand voice and policy skill

    Turns your tone of voice, your best past replies and your returns, delivery and warranty policies into one skill that every other customer service job loads first.

    Reads
    Your brand guidelines, saved replies and macros, and your published policies
    Who checks it
    The customer service lead approves the skill and owns every change to it.
  2. 02

    Reply drafts with the order already looked up

    Drafts the reply to each new ticket in your voice, with the order status, tracking and the policy that applies already looked up, so the agent starts from a finished draft.

    Reads
    Zendesk, Gorgias or Freshdesk; orders and customers in Shopify or your order system
    Who checks it
    An agent reads, edits and sends every reply. Nothing is sent automatically.
  3. 03

    Ticket triage and tagging

    Reads each incoming ticket, tags it by reason and urgency, and routes it to the right queue or person, with anything angry, legal or safety related flagged for a lead.

    Reads
    New tickets in your helpdesk; your tagging and routing rules
    Who checks it
    The lead reviews the flagged tickets and adjusts the rules when a tag is wrong.
  4. 04

    Refunds, returns and exchanges

    Checks each refund, return or exchange request against your policy and the order, and drafts the response and the next step, with the reason it qualifies or does not.

    Reads
    The ticket, the order in Shopify or your order system, and your returns policy
    Who checks it
    A person approves every refund and decides any exception. Claude never issues money.
  5. 05

    Knowledge base articles from solved tickets

    Reads the questions that keep coming up in solved tickets and drafts a help centre article or saved reply for each one, in your voice and from your policy.

    Reads
    Solved tickets in your helpdesk; your existing help centre
    Who checks it
    The customer service lead edits and publishes every article.
  6. 06

    The weekly themes report

    Writes a short weekly report on what customers contacted you about: the main reasons, what changed from last week, and the product, delivery or website issues worth passing on.

    Reads
    Tickets and tags from your helpdesk
    Who checks it
    The lead checks the themes against the tickets before sharing them with product, operations or marketing.
  7. 07

    Review and social message replies

    Drafts replies to new reviews and social messages in your voice, and moves anything that is really a complaint or an order problem into the helpdesk.

    Reads
    Google reviews, your review platform and your social inboxes
    Who checks it
    A person posts each reply. Anything negative or sensitive goes to the lead.

What the full framework adds

The PDF is about ten pages and free. It takes each automation above and adds a worked example of one typical run, the unit to time it in, a 30-day plan to prove your first one, the policy points for this team, a worksheet to measure before and after, and the mistakes we see most often.

  • Week 1Write the voice and the policies down
  • Week 2Pick one ticket type and time it
  • Week 3Build it and run it beside the old way
  • Week 4Decide, then pick the next job

See it in practice: NZ retail brand. A New Zealand retailer on Shopify and Microsoft 365, with seven departments running their day-to-day through Claude.

Want the whole method?

The AI Transformation Playbook covers the plan, the policy, the second brain and the skills for a whole business. Or build them with your team on a Claude training day.