The AI features are real.
Every major CS platform has shipped something in the last eighteen months. AI-generated health scores. Automated summaries. Predictive churn models. Sentiment analysis. Smart alerts.
If you've been watching your CS vendor's release notes, it looks like meaningful progress. And in some ways it is. These features are genuinely useful.
But there's something important to understand about what they are and what they aren't.
They're AI inside someone else's operating model.
The Container Problem
When a CS platform ships an AI feature, that feature is built on top of the platform's existing data model. It analyzes the data the platform has. It surfaces insights based on the signals the platform can detect. It operates within the workflow structure the platform was designed around.
Which means it's still constrained by all the same limitations the platform had before the AI features shipped.
If your CS platform only has access to structured data from your product and CRM, its AI features can only reason about that data. The call recording where your customer mentioned they were evaluating a competitor? The Slack message where a champion said their budget was being cut? The support ticket that revealed a deeper frustration than the surface-level issue? Unless those are in the platform's data model, the AI doesn't know they exist.
The intelligence is real. The container is still the container.
The Faster Horse Problem
There's a quote — probably misattributed to Henry Ford — about customers asking for faster horses instead of cars. The idea: incremental improvement within an existing paradigm is not the same as a paradigm shift.
AI features inside a CS platform are faster horses.
They're genuinely faster. The health score updates more intelligently. The alerts are smarter. The summaries save time. These are real improvements.
But the operating model is the same. You're still working within the schema the platform was designed around. Your operation is still conforming to their best practices. The ceiling is still the platform's ceiling — it's just slightly higher now.
A faster horse is still a horse. It can't do what a car does.
What a Paradigm Shift Looks Like
The paradigm shift is building an agentic layer that sits above all of your tools — not inside any of them.
Instead of your CS platform's AI reasoning about the data your CS platform has, you build workflows that pull from everything: your CS platform, your CRM, your product usage data, your call recordings, your support tickets, your unstructured context. All of it synthesized together, processed according to your business rules and your signal definitions, and delivered to your CSMs in a way that's actually useful for your specific operation.
The intelligence isn't constrained by what one platform can see. It's informed by everything you have access to.
And the workflow isn't constrained by the platform's schema. It's built around how you actually work.
That's the difference between AI inside someone else's operating model and AI that serves yours.
The LTV Implications
This matters for LTV because the signals that actually predict expansion and churn are usually not the ones that show up cleanly in structured platform data.
A customer who is about to expand doesn't just have a high health score. They've been talking about their growth plans on calls. Their champion just got promoted. Their usage of a specific feature has been trending up in a particular pattern. Their latest support ticket was about a capability they don't currently have.
None of that is fully captured in the data your CS platform's AI is reasoning about. All of it is accessible to an agentic workflow that pulls from every source you have.
The difference between catching an expansion signal at the right moment and missing it entirely is often the difference between a customer who expands before renewal and one who doesn't expand at all. That's an LTV gap. And it's one that AI inside a CS platform's container can't close — because it can't see the signals.
Your CS tool added AI. It's a real improvement. Use it.
And then build the agentic layer on top of it that does what the platform's AI was never designed to do.
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.