How to Build AI-First Customer Success Playbooks That Amplify Your CSMs

AI is changing the way customer success teams work, but the goal should not be to replace customer success managers. It should be to help them make better decisions, identify risks earlier, and spend more time building meaningful customer relationships.

In this panel episode of The Customer Success Pro Podcast, host Anika Zubair brings together Sareena Hirani of Amplitude, Elena Cocco of Alice QI, and Io Michels of Chattable to discuss how customer success teams can build AI-first playbooks that amplify their strongest CSMs. They share practical examples of using AI to identify renewal risks, capture expert judgment, improve customer intelligence, and make customer success processes more consistent.

The conversation also explores what happens when AI implementation goes wrong, why clean data and governance matter, and how leaders can introduce AI without losing the human skills that make customer success effective. If you want to make AI more useful for your CS team while protecting the customer experience, this episode is for you.

Building AI-First Playbooks That Amplify CSMs

AI-first customer success does not mean removing the customer success manager from the process. It means identifying the decisions and actions that make your strongest CSMs effective, then using AI to help more people apply those same practices.

Elena Cocco explains that building an effective playbook starts with understanding how experienced CSMs make decisions. It is not enough to simply document what they do. Leaders need to ask why they made a particular decision, what signals they noticed, and what influenced their next step.

These explanations help teams capture the judgment behind successful customer interactions. AI can then support repeatable actions and provide guidance, but its outputs still need to be checked against leadership expectations and actual customer success practices.

This is an ongoing process, not a one-time exercise. As customer needs and team behaviours change, playbooks should be reviewed, tested, and improved. The goal is to make expert knowledge easier to share without assuming that AI can automatically understand every situation.

Using Customer Intelligence to Identify Churn Risks Earlier

One of the most valuable applications of AI in customer success is identifying signals that might otherwise be missed.

Sareena Hirani shares how her team uses customer-call notes and AI skills to detect subtle renewal-risk indicators. These can include specific phrases, changes in tone, or patterns in customer conversations that may not be obvious when a CSM is reviewing an account with an overly optimistic or pessimistic perspective.

Amplitude's customer intelligence dashboard also brings together churn and expansion signals to help teams understand account health. However, having more data does not automatically mean better decisions. CSMs still need to understand what the signals mean, which ones matter, and where their attention can make the biggest difference.

This is especially important when customer success managers are responsible for large portfolios. Io Michels points out that some CSMs manage between 200 and 300 accounts. It is difficult to give every customer the same level of attention, so AI can help surface the accounts that need timely human intervention.

As Io explains, AI should help teams make better use of their limited time. It can highlight where attention is needed, but the CSM still needs to decide what action to take and how to approach the customer.

Making AI Adoption Practical Across the Team

Introducing AI across a customer success organisation requires more than giving employees access to new tools. Teams need time to experiment, learn from one another, and understand where AI can genuinely improve their workflows.

The panel discusses different approaches, including dedicated AI experimentation time, workshops, hackathons, and searchable libraries of successful customer plays. These approaches help teams share useful examples, learn from practical experiments, and make successful behaviours easier to repeat.

Another important consideration is where AI lives within the team's existing workflow. Io recommends bringing AI into the tools CSMs already use rather than expecting them to leave their usual work environment to access another platform.

This makes adoption easier because AI becomes part of the daily process instead of another task employees need to remember. When useful information and recommendations are available in familiar tools, CSMs are more likely to use them consistently.

The panel also emphasises starting with simple, repetitive tasks. As one of the key recommendations from the discussion puts it, start with the most annoying thing. Look for work that takes up unnecessary time, understand the process, and identify where AI can make it easier.

However, the process itself needs to work before AI is added. If a workflow is already broken, adding AI will not fix the underlying problem. It may simply make the same problem happen faster.

Learning From AI Failures and Keeping the Human Element

The panel is also honest about what has not worked. Examples include inaccurate AI-generated numbers, attempts to replace one-to-one meetings and quarterly business reviews, overly long AI-generated reports, and spending an entire year trying to create one perfect AI churn score.

These experiences highlight an important lesson: AI implementation should solve a real problem, not become a goal in itself.

AI can process information and surface patterns, but it cannot replace the human judgment needed to understand customer relationships. CSMs still need curiosity, ownership, prioritisation, business acumen, and strong conversation skills.

As one of the panel's key reminders puts it, "AI is not going to tell you what to do." It can speed up decision-making and help teams identify where to focus, but people remain responsible for interpreting the information and choosing the right response.

Good AI implementation also depends on reliable data and clear governance. Teams need shared definitions, appropriate access restrictions, cross-functional alignment, and a clearly identified owner for each AI-enabled process. Customer-facing AI agents require additional care and guardrails to ensure that automated interactions meet the organisation's standards.

The objective is not to automate every customer interaction. It is to remove unnecessary manual work so CSMs can spend more time on the conversations and decisions where their expertise matters most.

Key Takeaways

• Capture your best CSMs' expertise. Ask experienced team members why they make specific decisions and turn those insights into repeatable playbooks.

• Use AI to identify customer risks. Analyse call notes, customer behaviour, and account signals to identify potential churn and expansion opportunities earlier.

• Prioritise human attention. AI can help CSMs manage large account portfolios by highlighting where their time and effort can have the greatest impact.

• Bring AI into existing workflows. Make AI accessible through the tools CSMs already use instead of creating additional steps.

• Start with repetitive tasks. Focus on the most time-consuming or frustrating parts of the workflow, and fix broken processes before adding AI.

• Keep the human element. AI can support decision-making, but curiosity, business judgment, ownership, and customer relationships still depend on people.

• Establish clear data governance. Use reliable data, shared definitions, access controls, and accountable ownership to support responsible AI adoption.

The biggest lesson from this conversation is that successful AI adoption starts with understanding how your team works. Identify what your strongest CSMs do well, find the repetitive tasks that slow everyone down, and use AI to make good decisions easier to repeat.

Start with the basics, keep experimenting, and make sure your technology supports the people responsible for your customer relationships.

🎧 Listen on your favorite platform:

🔹 YouTube

🔹 Spotify

🔹 Apple Podcasts