Skip to content

All courses

Course 9 of The AI Decision Course

AI in customer service and experience

For leaders bringing AI into customer service: launching the first chatbot safely, who gets a person, an owner for every reply, letting AI help your live agents, and the number your service answers to.

13 short lessons. Each one is a short read with a card your team fills in and three AI prompts to test it. Part of The AI Decision Course: membership is $299 a year and includes all 12 courses, with a full refund within 14 days.

Join for $299 a yearRead the free lesson

About this course

This course covers bringing AI into customer service and the customer experience: launching a first chatbot safely, deciding who gets a person and who gets the AI, and giving every bot answer an owner. It also covers AI helping your live agents, personalization with restraint, reading and scoring every conversation, and picking the numbers your service answers to.

You leave able to launch a chatbot on a single question and widen it only once the number it was meant to move has moved, and to give every answer an owner and a review date. You can also sort your ticket queue before promising savings, put AI under your live agents a rung at a time, and pick a service number that counts whether the customer got help.

Lessons in this course

  1. Launch the chatbot on one question, because you answer for every reply2-minute read, Free to read
  2. Who gets a person, and who gets the AI3-minute read
  3. An owner for every answer the bot gives3-minute read
  4. Sort your hundred tickets before you promise the savings3-minute read
  5. Let AI help the live agent, one rung at a time3-minute read
  6. Personalize to one person, and know where to stop2-minute read
  7. Where conversation beats a form, a menu or a search box2-minute read
  8. Read every conversation, then put a date on the fixes3-minute read
  9. Let AI score every conversation, and keep it fair3-minute read
  10. Pick the number your service answers to3-minute read
  11. Choose the languages your AI serves, and to what standard4-minute read
  12. Reach the customer before they call, but only on a real signal3-minute read
  13. Test a voice agent on real calls, not transcripts3-minute read

Back to the course list

Questions

How do we launch a customer service chatbot safely?

Start with the question customers ask most and answer it correctly, then add the next question only once the number it was meant to move has moved. You answer for every reply the chatbot gives, so tune the key answers with the people who know them before launch. The lesson 'Launch the chatbot on one question, because you answer for every reply' covers the approach.

When should a customer get a person instead of AI?

Route high-stakes and upset callers to a person, and keep an easy route out of anything you automate. Offer the AI as a choice rather than a default, and let callers change their mind at any time. The lesson 'Who gets a person, and who gets the AI' covers the routing.

Should our support team track deflection or resolution?

Deflection counts contacts that never reached a human, not whether they got help, so run the service on resolution instead. Verify resolution against the conversations before counting it, and pair every cost number with an outcome. The lesson 'Pick the number your service answers to' covers the choice.

The other courses in the membership