I got caught out by AI this week.

It got me thinking about the impact on our customers if the same happens to them.

It’s a warning, but I will also provide you with some practical steps further down!

So, let's talk about AI-generated responses for a moment.

I experienced this on one of my recent LinkedIn posts.

Can you spot the AI?

When I saw the first comment, I felt a brief flutter of appreciation. It was nice to have validation that people are getting value from this free content I create each week about customer onboarding.

But then…

Another comment came in and it was eerily similar in structure and tone.

That initial feeling of connection evaporated into disappointment as I realised these weren't genuine human interactions at all.

This exact emotional journey, from appreciation to disappointment, is what happens when companies over-automate their customer onboarding.

It's a graph of diminishing trust that none of us want our customers to experience.

If this happens on something as simple as LinkedIn comments, imagine the impact on customer relationships during onboarding.

So, how do we utilise technology like AI and LLMs (Large Language Models) to assist our onboarding experiences rather than hinder them?

I’m glad you asked, keep reading. ↓

🧪 The Project: Human-First AI in Onboarding

This is a practical guide with plenty of examples to inspire you.

It’s worth checking these steps to make sure you don’t inadvertently jeopardise your relationship with your customer.

Step 1: Identify high-touch points

When you map out your customer journey for onboarding we often focus on the milestones, the tasks, and responsibilities.

I suggest adding an extra layer that identifies the high-touch human points. This helps to determine:

• Critical moments that require human expertise and empathy

• Situations where customers need reassurance or complex problem solving

• Key relationship building opportunities

• Points where misunderstandings could lead to customer frustration

A well timed human conversation can prevent weeks of back and forth with automated systems.

The aim is to enhance rather than diminish the customer experience through protecting these vital human touch points.

Step 2: Automate the right tasks

Where AI can come into it’s own is through repetitive, time-consuming activities that don't require human judgement.

Many of these will be internal processes, which reduces the risk of direct impact to the customer if things don’t go to plan.

Examples:

• Document collection and organisation

• Basic FAQ responses and resource sharing

• Meeting scheduling and follow-up reminders

• Progress tracking and milestone notifications

Consider being transparent - remember this previous project? ↓

Step 3: Create smart escalation triggers

AI still has a long way to go to be a one and done process. It requires constant iteration. It can create more work than it’s worth, if you don’t have a team to focus on it.

That’s why it is important to develop clear signals for when AI should hand off to humans.

Example triggers:

• Customer expressing frustration or confusion

• Multiple repeated questions about the same topic

• Complex technical issues or unique use cases

• High value accounts or strategic opportunities

Think of these triggers as your safety net. They can ensure that customers never feel stuck in an AI loop when they need human assistance.

This simple step may well save your onboarding.

Step 4: Blend AI efficiency with human expertise

The crux of my opinion on this topic is to use AI to enhance, not replace, human interactions.

We can use it as a powerful tool to become more efficient and give us more opportunity to provide even better service to our customers.

Some examples of how this can work:

• Let AI prepare customer insights for human focused calls

• Use predictive analytics to help prioritise tasks

• Automate meeting summaries and follow up tasks

• Enable smart note taking during customer conversations

🤓 The Analysis

Our customers are not AI bots… yet.

We haven’t reached the inception point where AI customers deal with AI CSMs. What a weird day that’ll be.

Right now, AI can't replicate the nuanced understanding that comes from years of customer onboarding experience.

It can't read between the lines of a customer's hesitation, or spot opportunities to expand value that weren't explicitly mentioned.

Most importantly, it can't build the kind of trust that keeps customers loyal for years to come.

Stay human, stay authentic, and let AI handle the busy work while your team focuses on what they do best - creating successful, satisfied customers.

What to expect if you try this project:

• Much deeper customer relationships & trust

• More time for meaningful customer conversations

• Higher customer satisfaction & retention rates