The New CSM Workflow: What AI Handles vs. What You Still Own

AI can handle the work. But it still can’t replace you... if you focus on the right things.

I was catching up with a VP of Customer Success recently. Her team manages hundreds of accounts - large, complex, high-stakes relationships.

We were talking about how the team’s working, what’s working, and what’s not. At one point, she said something that stuck with me:

“We’ve got great people. But most of them are still spending their days doing work AI can already do.”

She wasn’t frustrated. She was worried. Because these are smart, capable CSMs. But their day-to-day hasn’t evolved much. Meanwhile, the tools around them have.

And the reality is that her top performers aren’t necessarily the ones with the most experience. They’re the ones who’ve adapted. They’ve figured out where they still make the biggest difference and where it’s time to let the machines take over.

If you’re in CS right now, this is the shift.

It’s not about losing your job to AI. It’s about realizing the job is changing.

And figuring out how to stay ahead of it.

Let’s talk about how.

What AI Can Now Handle (no…not the usual suspects)

We’ve all seen the obvious stuff. AI can write your meeting recaps. It can draft emails. It can help build your QBR deck. That’s table stakes now.

But the CSMs I talk to… the ones staying ahead are using AI for much more interesting, strategic things. Not just to save time, but to think more clearly, prep more deeply, and move faster than they could before.

Here are four ways I’ve seen it used recently that really stood out.

1. Practice the conversation before it happens

One CSM was prepping for a high-stakes renewal. Her champion was about to pitch the deal internally, and she wanted to help them prep.

So she dropped the internal deck into GPT and asked:

“If I were the CFO seeing this deck, what questions or concerns would I have based on [x] industry?”

GPT came back with five tough, but fair objections. She coached her champion through each one. The deal closed two weeks ahead of schedule.

Most CSMs wait until a customer says no to hear their concerns. But in this case, the CSM saw them coming and defused the risk ahead of time. As a CSM this is golden. And as a CS leader - this is what you call out in your next senior leadership meeting / board. Don’t forget to make sure the CSM gets their due recognition.

2. Turn vague goals into real plans

You’ve probably heard a customer say something like:

“We just want to improve onboarding.”

Or “We need to see more value.”

But what does that actually mean?

One CSM took the goal of, “reduce time to value for new users” and asked GPT to break it down.

How should Product contribute? Support? CS? Enablement?

GPT gave her a full breakdown by function. She turned that into an internal one-pager, shared it with her team, and then built a customer-facing success plan from it.

Now they weren’t just nodding along to a vague goal. They were actually doing something with it.

3. Spot competitive risk before it hits

Another CSM I spoke to had set up a simple GPT-powered agent to monitor competitor mentions inside her accounts. Like LinkedIn posts, job listings, team updates.

One day, it flagged a job posting from a key customer that said, “Experience with [competitor] preferred.”

She brought it up with the customer directly. Turns out, they were quietly evaluating other vendors. That conversation wouldn’t have happened if she hadn’t been watching.

You can’t rely on your champion to tell you when things are shifting.

Sometimes, you need to read between the lines or get AI to do it for you.

4. Get your internal escalations heard

When customers are blocked, most CSMs Slack product or submit a ticket.

But one CSM took a different approach. She gave GPT all the context:

• ARR of the account

• The blocker

• Quotes from the last call

• Support history

Then she asked GPT to write a one-paragraph summary for the product team - something clear, urgent, and specific.

That message landed. The fix was prioritized. The problem got eventually solved.

We often assume other teams are ignoring us. In reality, they just don’t have the full picture. This CSM made it easy to care and easy to act.

What you still own

So yes, AI is getting better at tasks.

But it’s still not great at judgment. Or trust. Or just…nuance.

That part’s still on you.

Here’s where I’m hearing the best CSMs stand out.

Reading the room is still a thing..

AI doesn’t notice when someone goes quiet on a call.

Or when your champion stops CC’ing their boss.

You do. And those signals matter. I’ve talked about this before in a previous write up.

Asking better questions

AI can tell you what a customer asked for. But…only you can figure out what they actually need.

One CSM told me a customer requested a “Salesforce-style report.”

Instead of logging it as a feature request, she asked why. Turned out, they just needed a board-ready export.

She showed them a quick workaround. No product work required. Problem solved.

Having hard conversations

When something breaks, or a launch slips, or a feature’s not ready AI’s not going to hop on that call for you. And even if it could… who wants to talk to a bot?

Those moments require trust. And trust doesn’t get built with a prompt. It gets built with consistency, presence, and honesty over time.

Leading change

Most adoption problems aren’t about product. They’re about people. And the way people work.

Helping a customer navigate internal resistance, retrain teams, or shift priorities? That’s human work. And it still matters. My theory is that change management will always be human-led. It requires empathy, observation, patience, and awareness. Things that, in my experience, AI still hasn’t mastered. Not in any real, relational way.

You need to know when a team is resisting change not because the product is confusing.. but because they’re scared it’s going to change their role. Or threaten their status. Or make their job harder in the short term.

That kind of insight doesn’t show up in a dashboard.

It shows up in a side comment, a tone shift, a throwaway line at the end of a call.

And when you catch it, and respond the right way, that’s what creates real momentum. That’s what moves customers forward.

Advocating internally

Sometimes the thing your customer needs doesn’t scale.

And someone has to make the case for why it matters anyway.You’re not just a messenger. You’re an advocate. That work still requires influence, not automation.

What to do next

This isn’t about working more. It’s about shifting your focus.

Start by looking at your week. What are you doing that a tool could already do faster or better?

And what are you doing that still requires your judgment, your presence, your voice?

The best CSMs I know are asking themselves that question often.

They’re not afraid of AI. They’re learning to use it well, so they can spend more time where they actually add value.

Let AI take the task.

You take the outcome.

That’s how you stay ahead.