AI in Customer Success: The Loudest Conversation, the Least Grounded

AI was supposed to make go-to-market teams faster, smarter, and more effective. In customer success, it was supposed to unlock proactive engagement, better customer experiences, and more strategic relationships.

But somewhere along the way, something went sideways.

In a recent episode of PROMPTED, I sat down with Mike Lemire, a longtime customer success leader and advisor, to talk candidly about what's actually happening inside CS organizations right now. 

The short version: AI strategy has become the loudest conversation in customer success, and often the least grounded.

When "AI Strategy" Becomes the Only Strategy

Mike works with CS leaders across companies of all sizes, and he sees the same pattern repeating. Five years ago, every CS conversation eventually led to the same question: should we buy Gainsight? Today, that question has been replaced with a new one: What's your AI strategy?

On the surface, that sounds like progress. In practice, it often is not.

Instead of starting with problems, teams are starting with tools. Instead of asking how AI could meaningfully improve the customer experience, they're asking which AI product they should buy so they can say they're “doing AI.”

As Mike put it during the conversation, the pressure rarely comes from customers. It comes from leadership teams and boards that want to see AI show up in decks and quarterly updates. The result is AI adoption driven by optics rather than outcomes.

That shift matters, because it changes how success is measured. The question becomes whether AI is being used, not whether it's helping.

The Quiet Incentive Behind the Push

There's another layer underneath this trend, and it's not talked about very often.

Investor pressure plays a role. AI is the latest technological wave, following cloud and mobile, and capital has already placed its bets. That creates a sense of urgency and fear of missing out at the leadership level. If everyone else is “doing AI,” then not doing it feels like falling behind.

At the same time, AI is frequently framed as a margin lever. Reduce headcount. Increase account loads. Do more with fewer people.

What's missing from many of these conversations is clarity about what “more strategic work” actually means in customer success. In reality, many teams are not using AI to deepen relationships or deliver more value. They're using it to stretch already overloaded teams even thinner.

Support As the AI Testbed

Nowhere is this more visible than in Support.

Support has quietly become the proving ground for AI, largely because the workflows are repeatable and the data is plentiful. Chatbots, knowledge bases, and automated responses are easy places to start.

Some of that progress is real. Modern support bots are better at understanding context and natural language than their earlier if-then predecessors. In many cases, they do help customers get answers faster.

But Mike pointed out a subtle and troubling shift in language. Teams are moving away from talking about ticket resolution and toward talking about ticket deflection.

Deflection sounds efficient; but resolution is customer-centric. The difference matters.

When success is measured by how many interactions are avoided rather than how many problems are solved, the customer experience slowly erodes. Builders and leaders alike should pay attention to which metrics they're optimizing for, because language reveals incentives.

The “Customer Watermelon” Problem

One of the most useful mental models Mike shared was the idea of the “customer watermelon.”

A customer can look green on the outside and red on the inside. Positive sentiment. Friendly conversations. Strong relationships. But very little product usage or value realization.

At the same time, a demanding or frustrated customer may be deriving enormous value from the product and would never consider leaving.

AI tools make sentiment easy to measure, which is exactly why it becomes dangerous when overweighted. Sentiment is convenient. Value is harder to quantify.

Customer success has always been about value realization. AI doesn't change that; it just makes the tradeoffs more visible.

A Hackathon That Changed the Build Vs. Buy Equation

One of the most compelling stories from the conversation involved a CS team debating whether to invest in Gainsight. Instead of buying too early, they ran a hackathon.

CSMs were given AI building credits and asked to recreate only the features they actually cared about. Not a full platform, just the workflows that mattered most. An engineer helped clean things up at the end.

The result was a messy, imperfect internal tool that worked well enough to run for months. More importantly, it taught the team what they actually needed, how their data flowed, and when they would truly be ready to invest in a bigger platform.

As Mike put it bluntly, vibe coding is already taking six figure deals off the table today. Not someday. Today.

This is a massive signal for both GTM leaders and builders. AI has lowered the cost of experimentation to the point where internal tools can meaningfully delay or reshape enterprise purchasing decisions.

What This Means for GTM Leaders

The biggest takeaway from this conversation is not that AI is bad or overhyped. It's that intent matters more than technology. AI works best in customer success when it does three things well:

• Removes administrative friction so humans can be more present with customers
• Improves decision making about where time and energy should be spent
• Strengthens the feedback loop between customers, CS, and product teams

What it should not do is replace authenticity, hide customers from humans, or become a proxy for strategy. Consider this an AI north star: If your customer were in the room when you explained how you plan to use AI, would they feel better about the experience they're about to have?