How AI Is Changing the CSM Role: From Knowledge to Context

Ask most CSMs what makes them good at their job, and they’ll tell you some version of the same thing: they know their stuff. They’re well-versed in the product, its use cases, industry benchmarks, and workarounds. That knowledge took time to build, but AI has made it table stakes.

This doesn’t mean CSMs are less important. It just means the job has changed. Now that anyone can access best practices and product documentation in seconds, simply knowing things isn’t what sets you apart. What matters is using that information to do something specific, timely, and truly helpful for each customer. This shift from knowledge to context is changing what great Customer Success (CS) looks like.

Nowhere was this clearer than in a recent episode of the [Un]Churned podcast, where Diane Wu, Global Head of Customer Success and Experience at Google Cloud Security, and Brady Bluhm, Senior Product Manager for Staircase AI at Gainsight, got into what this shift looks like on the ground and what CS leaders should actually do about it.

What Does Context-Driven Customer Success Actually Look Like?

In practice, this means your CSMs spend less time collecting information and more time figuring out how to use it.

Consider the usual pre-meeting prep: open the account in the CRM, check recent emails, try to recall the last call, and look for the latest health score. This process is fragmented, manual, and relies on one person’s memory. This is what AI is intended to replace. The most successful teams now treat customer context as something everyone in the organization can access and build on, not just something individual CSMs keep in their heads.

This means building shared customer repositories where every call transcript, every slide deck, every action item gets captured in one searchable place. Diane’s team at Google Cloud Security uses NotebookLM to do exactly this. It’s a living record for each customer that product managers and engineers can access just as easily as the CSM can. As she put it, it creates “highly productive CSMs” and “highly productive internal communication on customer context.” The wall between what CS knows about an account and what the rest of the company acts on starts to come down.

Although AI handles the synthesis, it’s the CSM’s job to know what questions to ask, how to pressure-test the output, and how to turn an AI-generated summary into something a customer actually finds useful. In Diane’s framing, the CSM needs “intellectual curiosity and technical fluency to prompt it the right way.” Judgment, relationship awareness, and the ability to spot when something’s off aren’t going anywhere. If anything, it matters more now that the information-gathering is handled.

Why Do Top-Performing CSMs Resist AI Adoption?

Many CS leaders assume their best people will adopt new AI tools the fastest, but Diane learned firsthand that this isn’t always true.

When she introduced NotebookLM to her team, she thought her top performers would embrace it right away. “What I found is actually it was almost the opposite,” she shared. Her best CSMs had spent years building their own systems, personal workflows, and habits that produced results. Asking them to change meant asking them to move away from something that already worked.

Newer CSMs don’t have that issue. They don’t have established routines to protect, so a tool like NotebookLM doesn’t feel disruptive. In fact, it probably feels like exactly what they needed. They adopt it quickly, see results fast, and move on.

So what does this mean for CS leaders trying to encourage AI adoption? Plan your rollout based on how disruptive the change will be, not just on skill level. The best entry points are those that remove friction from tasks your team already finds difficult, like call prep, EBR creation, or updating CTAs. Diane calls this the “wow moment”—a small, clear win that changes how someone feels about a tool. Start with Gemini notes and transcripts before moving to more advanced workflows. Give people that moment, and curiosity will take over.

How Is AI Changing CSM Coverage Models?

The traditional coverage ratio, like one CSM to ten enterprise accounts or one to 20 in a more efficient model, was based on certain assumptions about what a person can do in a workday. This includes how long it takes to research an account, prepare for a QBR, and how much time is spent on tasks that aren’t actual customer conversations.

AI changes those assumptions, so it’s time to rethink the ratios. When a CSM can pull a pre-meeting brief in minutes instead of an hour, run five customer analyses in the time it used to take to run one, and get a full account history in seconds without searching through old emails, the math on what one person can handle changes a lot.

However, this shift shouldn’t open up the floodgates on CSMs and require them to take on more accounts. Instead, coverage model decisions should reflect what AI-assisted workflows actually make possible, not what manual workflows require.

What stays the same through all of this is the human at the center of the model. Memory used to be a key trait of a great CSM, and their ability to recall every conversation, commitment, and detail of a complex account was part of what made them great. AI takes care of that now. The CSM is no longer the person who remembers everything, but the one who knows what to do with that information: asking better questions, reading the room, building relationships, and turning insights into results that matter to the customer’s business.

What Does AI-First Customer Success Look Like in Practice?

Most CSMs today work across more tabs than they’d like to admit. CRM is in one place, health scores in another, call recordings somewhere else, and the AI tool in a different window. Each switch takes a little focus and time. Over a day, these add up, which is why even well-resourced CS teams can feel stretched.

The next wave of AI in post-sales is less about adding smarter individual tools and more about connecting the ones you already have. MCP (Model Context Protocol) integrations let CSMs interact with their entire tech stack using natural language. They can pull account history, update health scores, find open CTAs, and draft follow-ups without leaving their main workspace. Diane sees this as a big shift in how CS professionals will work: “The traditional application UI is going to eventually fade and go away, and LLMs will become the new workspace.” The difference, as Brady saw when testing Gainsight’s Staircase MCP server, is a game-changer. He surfaced and closed CTAs that had been open for two years with a single prompt, without logging into the platform.

The downstream benefit is the ability to deliver customer engagement that actually feels personalized, even when a single CSM covers a portfolio that would’ve required a much bigger team to manage a few years ago. That’s the version of Customer Success that protects revenue, creates expansion opportunities, and earns CS a permanent seat at the table.

The informational edge is gone. But the human edge, with our judgment, relationships, and the ability to turn context into action, has never been more valuable. The teams building systems to support this are the ones to watch.

Want more like this?

Every week on the [Un]Churned podcast, host Josh Schachter talks with post-sales leaders working on retention, expansion, and adaptation to the agentic era. Listen to the full episode with Diane Wu of Google Cloud Security and Brady Bluhm of Gainsight here, and subscribe to the [Un]Churned Substack for weekly breakdowns delivered to your inbox.