Quick summary: Autonomous customer success isn’t the end of human connection in CS; it’s actually quite the opposite, says ChurnZero’s Abby Hammer. Here’s how to make it work.
Abby Hammer holds two roles that rarely share a seat at the same table: chief product officer and chief customer officer at ChurnZero.
That combination is intentional. The best product decisions, she says, come from being deeply immersed in the day-to-day realities of the teams you serve.
On a recent episode of The Growth Signal podcast, Abby sat down with host Alyssa Nolte to talk about where CS is heading and why autonomous customer success is closer than most teams think.
Read on for seven insights from Abby, in her own words,on:
- How AI is closing the gap between managing relationships and managing admin.
- Why AI isn’t damaging but enhancing the experience of SMB and long-tail customers.
- Why it’s okay to be uncomfortable with old skills losing value, and how to reframe it.
Plus… the single most common AI implementation mistake you can make and how to avoid it. Let’s dive in.
1. Autonomous customer success doesn’t mean impersonal.
AI and automation are going to run the majority of customer relationships… and I genuinely believe it will allow customer success to be more human than it’s ever been.
CS has always had a bit of an identity crisis. We say the role is about relationships and outcomes and strategy, but in practice, teams spend a lot of their time drowning in the admin: logging notes, making decks, chasing data.
Automation has been chipping away at that for years through onboarding sequences, health scoring, alerts, and renewal triggers. There is real ‘plumbing’ in place. What AI adds is the brain.
It’s not just “Hey, send that email at day 30.” It’s: “Here’s what this customer needs right now. Here’s a way to say it that’s going to resonate with this particular group.” Automation handles the motion. AI handles the judgment.
This plays out differently across segments. For high-touch teams, CSMs get freed from operational noise so they can focus on work that actually matters. For lower-touch and SMB segments, it raises the floor entirely. Those customers haven’t been able to receive meaningful, human-forward service because it just hasn’t been scalable.
This isn’t AI replacing something customers already had; it’s the ability to deliver something they’ve never had before.
2. CS is supposed to be about relationships. AI takes us back to that.
No one got into customer work to fill out notes and update fields in a CSP or CRM. People are attracted to the role because they like forming relationships. They like having an impact and getting to see that. That is the kernel of truth customer success was built on.
However, if we’re truly honest, for a lot of us, the last 15 years have actually been more about customer administration.
Yet, no one renews because they think their CSM logs great notes about them. They renew because they have a CSM who shows up in their conversations, adds value, and helps them rethink what they’re doing as a real thought partner. That’s only possible if we get rid of the noise.
This is why I don’t see this as the death of CSMs. Rather, it’s a removal of everything that’s diluted and distracted us from the role—and it’s giving CS teams a realistic and meaningful way to spend their time and energy on the work that lights them up inside.
3. When used well, AI doesn’t degrade the customer experience.
AI can make the customer experience feel more tailored, more timely, and more frictionless. It can keep up with customers in real time in a way that’s just not possible for a human—even a super well-intentioned, on-the-ball human.
Think about a customer you’ve had for two or three years. Think about everything that’s gone on in that relationship. Even if you wanted to show up knowing all of that, what does prep look like without something that helps you sort through what matters, understand the key points, and act as another CSM asking, what should I focus on? What’s going to move the needle?
Customers already expect you to know them. That’s table stakes. AI is what makes that expectation realistic at scale.
4. Letting go of old skills is uncomfortable, but leads to something better.
I use AI every single day and, in some ways, I feel a sense of loss around certain skills I used to identify as things that made me good at my job. Writing a great email, for example, used to move people forward in their careers. Now it’s table stakes.
So, instead of being sad about letting go of that skill, the question I ask is: what does this open me up to do?
I can’t afford a human assistant, but I can afford an AI one, so I welcome it taking these things. Now, my uniquely human brain can really focus on what matters.
The CSMs and CS leaders who will do well in this transition are the ones who lean in and ask that same question.
5. The biggest AI implementation mistake? Trying to feed it everything.
Too many CS teams try to give their AI everything—every metric, every data point—because if it has everything, it can’t possibly miss something.
Instead, your best bet is to choose one use case and build a clean strategy around it.
For example, if you want to stop taking notes during meetings, you don’t start by handing over your web analytics. You get a note taker, you build a transcription process, you clean the output, and you connect it to your CRM. That’s how you get a working solution. It’s not sexy but it works.
Context is everything. The right context at the right depth in the right place is where AI goes from a great idea to something that genuinely changes how a team works. When you get the fundamentals right, you open up the next layer of possibilities.
6. Imperfect data isn’t a dealbreaker, but you need a plan.
No technology, including AI, can fix a broken customer understanding.
If you have no idea what’s going on with your customers and you have no data, nothing is going to help you. Data still matters, and having it centralized still matters.
That said, you can actually use AI to solve your data quality challenges to a certain degree.
Some of the first AI agents we’ve introduced at ChurnZero are about data enrichment: keeping roles updated on accounts, reading engagements, or understanding the network of contacts at a customer’s company.
These are really important if you’re trying to understand customer dynamics or trigger automation that hits the right person at the right time—and the models are getting better and better at handling data that isn’t perfect.
7. If you’re not equipped for autonomous CS yet, find a good partner.
At ChurnZero, we are our customers’ platform, and we also take our role as their partner on this strategic shift really seriously.
We spend a ton of time thinking about how AI can be most useful, and two main themes stand out.
One—and this sounds simple—is where it lives. How much time are you going to waste copying and pasting stuff from a chatbot, for instance, to get the proper information to do something? When AI is naturally integrated into the workflow, we see adoption grow much stronger than when teams have to figure it out.
The second thing is context; a word that doesn’t feel big enough, because it’s not just context on the customer, though that is certainly important.
It’s how you put the proper context around the business you’re serving, the particular customer you’re talking to, the products and solutions you have, how that customer wants to be talked to, or how that CSM actually talks. When you think about all the things that go into making a truly great customer experience, the context is so deep.
That depth of integration is what allows AI to feel like a true teammate, as opposed to something you can obviously tell is machine generated.
It reminds me of when the Internet of Things was a big deal; I need to connect my light switches to my GPS so that when I drive into my driveway all the lights turn on. This is the next generation of everything needing to talk to everything else in the same language to take full advantage of the brain that AI is. One of the things we really have to grapple with is the level of deep thinking required to make it truly sing.
There are a lot of solutions out there that allow you to integrate your data as it is, enrich it, configure it for optimal AI consumption, then take action. If that is not something your organization is equipped to do internally, which most are not right now, find a good partner to do that with.
It’ll make the difference between getting disillusioned with the entire concept, versus having a good base to build on and keep exploring.
——-
Thank you to Alyssa Nolte for hosting Abby on The Growth Signal, the podcast for candid conversations on how customer relationships are built today: what’s working, what’s outdated, and what’s just weird enough to work. Listen to more of The Growth Signal interviews with post-sale leaders.
