Three layers of automation

  1. Macros — canned responses an agent applies with one click. The human stays in the loop and decides.
  2. Triggers — rules that fire on events: auto-assign by topic, auto-tag, escalate on SLA risk. No human action needed for the routing itself.
  3. Bots / AI replies — fully automated responses sent without an agent. Powerful and dangerous.

Start with macros, add triggers, and be very deliberate about the third layer.

Macros done right

  • Personalize the opening line. A macro that starts "Hi {{customer.first_name}}" beats a generic blast.
  • Keep them short. A macro should be a starting point an agent edits, not a wall of text they send blind.
  • Audit usage monthly. Retire macros nobody uses; split overloaded ones.

Triggers that earn their keep

  • Auto-assign by detected topic so tickets land with the right team.
  • Auto-escalate when a ticket is 30 minutes from SLA breach.
  • Auto-reopen if a customer replies to a "solved" ticket.
  • Tag by sentiment so angry tickets surface to a lead.

Where automation goes wrong

The fastest way to make a frustrated customer furious is to answer their carefully-written message with an obviously canned reply that ignores what they said.

Never auto-send when:

  • The sentiment is negative.
  • The ticket is a P1/P2.
  • The customer has already replied to a bot once and is clearly stuck.

The rule of thumb

Automate the routing and the busywork. Keep a human on the judgment and the empathy. When in doubt, let automation tee up the answer and let a person send it.