Everyone's building automation that removes people from the loop. The better play is automation that makes the people in the loop perform at their best — and when you're engineering for LTV, that distinction is everything.
There's a version of "AI-powered CS" that's becoming a cliché: the fully automated sequence, the AI that handles objections, the system that closes expansions while your team sleeps. I get the appeal. I also think most of it is solving the wrong problem.
The expansions that move LTV — the ones that stick, that compound, that deepen the customer relationship — still happen in conversations. A human calling another human, with something real to say, delivered at the right moment. That hasn't changed and isn't going to.
What has changed is everything around that conversation. And that's where the interesting work is.
The Prep Problem
Here's what kills the expansion moment before it starts: the person making the call doesn't have what they need. They're not sure what to lead with. They haven't done the math on what the customer's situation actually looks like. They don't know what the customer already knows. So the call is tentative. The energy is off. The customer picks up on it and the window closes.
Most teams respond to this by adding training. More role-plays, more scripts, more certifications. Some of that helps. But it doesn't solve the real problem, which is that every customer conversation is different and the rep is being asked to figure out the specifics on the fly.
The better answer is to eliminate the prep problem entirely. Before the rep picks up the phone, they have everything: the right opening based on what the customer does and doesn't know, the ROI math already calculated, the follow-up email ready to send. They walk into the conversation confident because there's nothing to be uncertain about.
That's a workflow problem, not a training problem. And it's a solvable one.
The measure of a good agentic workflow isn't how much it automates. It's how well it sets up the conversation that actually closes — and how much LTV that conversation generates.
Building Backward From the Conversation
The right way to design these systems is to start at the conversation and work backward. What does the rep need to walk into that call confident? That's the output the workflow has to produce.
For an expansion motion triggered by a specific customer milestone, that meant: a script matched to whether the customer knows what's been running or not, the pipeline math already done, a follow-up email ready to send the second the call ends. All of it waiting in a Slack notification when the trigger fires.
The rep's job isn't to figure any of that out. Their job is to make the call.
This sounds simple. The discipline is actually building the workflow that way — not adding features, not automating the parts that are easier to automate, but relentlessly focusing on what the person in the conversation needs and working backward from there.
Friction Is the Enemy
After the expansion call, the rep needs to confirm it was done so the next step — AE notification, same-day follow-through — can happen. The first instinct is always some kind of form or checkbox. Mark it complete, trigger fires, done.
Kill that idea. If the rep has to switch contexts to complete a step, they'll skip it. Not because they're lazy — because they're handling seven other things and every context switch has a cost. The AE never gets notified. The close doesn't happen. Another expansion event dies in the handoff.
The right design keeps everything in the same thread. The system pings the rep where they're already working. They reply yes. That's the trigger. No form, no tab-switch, no friction.
If they don't reply: reminder four business hours later. Still nothing: another four hours and their manager sees it. Two chances before it escalates. The whole path is baked in because every link in the chain matters when you're engineering for LTV.
Every automation decision serves a conversation between two humans. The tech is load-bearing infrastructure for a human interaction — not a substitute for it.
What This Isn't
This isn't a story about AI replacing your team. The rep still makes the call. The AE still closes the deal. The nuanced read on the customer's energy, the objections that came up, what the AE needs to know before they dial — that's irreplaceable. No workflow generates that.
The workflow's job is to make sure the right person has the right information at the right moment, and that nothing falls through the cracks between the trigger and the close. Every dropped expansion is a permanent LTV loss. The workflow exists to make sure that doesn't happen.
When you're designing these systems, the question to keep asking is: what does this person need to walk into this conversation with complete confidence? Work backward from that. The automation fills in everything except the conversation itself.
Get that right and the human moment takes care of itself.
Lincoln Murphy formally named and popularized Customer Success starting in 2010 and has spent 15 years connecting it to expansion revenue and commercial outcomes. Read The Premise.