Preparing to implement an advanced AI agent – Zendesk

Add-on

AI agents - Advanced

This article walks you through a successful implementation project for AI agents - Advanced. A solid implementation establishes a strong foundation for a well-performing AI agent.

This article contains the following sections:

Assembling your team
Setting goals
Collecting data and insights

Assembling your team

We recommended you include the following people on your AI agent implementation team.

Project lead

Their job title is typically something like Head of Customer Support, Head of Customer Experience, or Director of Customer Service and Operations.

The project lead is responsible for:

Owning the overall project
Having internal accountability for its success
Selecting and briefing the project team
Internally communicating strategic objectives
Remaining aware of progress and informed of any blocks

AI agent builders

Their job title is typically something like Customer Support Agent, Customer Service Agent, or Customer Care Agent. Choose two or three experienced people.

The AI agent builders are responsible for:

Understanding current processes and customers
Owning the AI agent building activities
Owning post-launch monitoring and content maintenance

CRM admin

Their job title is typically something like CRM Administrator, CRM Manager, Ops Manager, or Developer/Product Manager for API Integrations.

The CRM admin is responsible for:

Owning the CRM, widget, and technical setup
Participating in the kickoff and initial setup
Participating in testing and launch

Custom integration engineer

This role is needed only if the project includes an integration. Their job title is typically something like Backend Engineer.

The custom integration engineer is responsible for:

Knowing the data in the relevant backend systems
Understanding the use cases at a general level
Implementing backend integrations

Setting goals

It's important for the team to understand and agree what problems you’re solving and what you want to achieve.

Some example goals and KPIs are:

Reach a target custom resolution rate within one month of launch
Reduce average handling time (AHT) by a target percentage
Reduce first response time (FRT) by a target percentage
Increase the number of customers served through 24/7 support

Collecting data and insights

The following information is helpful to have for onboarding:

Link to your help center
Customer query report sorted by frequency or volume
Customer messages from a relevant timeframe
Agent training materials
Current agent workflows for common inquiries
Ideal AI-assisted workflows for common inquiries

Brand guidelines

Get a copy of your internal brand guidelines that describe the tone of voice to use when communicating with your customers. Often the marketing team owns this document.

Custom integration

If the project includes an integration, collect the following as well:

Agreed prioritization of the automations your team wants to build, as they may impact different backend integrations
Outputs from custom integration preparations