Welcome to Profit Loop AI, I’m adding a 2nd version of our newsletter focused exclusively on who AI is shaping the CS profession. 

This will be a weekly briefing on AI, agents, prompts and the future of Customer Success. 

I’ve been thinking a lot about where AI is actually taking Customer Success.

Not the hype. Not another article about how ChatGPT can write your emails. And not another list of 47 prompts that nobody will ever use.

Something much bigger is happening.

AI is moving from helping CSMs do their work to actually doing parts of the work.

And I think that changes Customer Success more than almost anything we’ve seen in the last decade.

This is the first edition of a new weekly series I’m experimenting with.

Every week, I’ll be digging into what’s happening at the intersection of:

I’ll filter out as much of the noise as possible and focus on one question:

“So what can a Customer Success professional actually do with this?”

Let’s get into it.

Most of us started using AI as an assistant. We ask it a question. We give it some information. It gives us an answer. We write the prompt. We review the output. We decide what to do next.

That’s useful.

But it’s still fundamentally human-directed.

Agents are different.

Instead of saying: “Analyze this account.”

You can increasingly say: “Every Monday, review my accounts. Look at usage, support activity, recent conversations, renewal dates and stakeholder changes. Identify accounts that have materially changed. Explain why. Recommend what I should do next and prepare the necessary follow-up.”

That’s a fundamentally different workflow.

The agent isn’t just answering a question.

It’s watching.

It’s reasoning.

It’s taking action.

And increasingly, it’s doing that across multiple systems.

Salesforce recently expanded Agentforce with agents designed to pursue goals over days and weeks, learn new skills, work together and continuously improve. Zendesk has similarly introduced specialized agents designed around specific business processes and connected systems.

This is the direction things are moving.

Forget building some complicated “AI CSM” that tries to replace the entire role.

Start much smaller.

I’d build a Customer Intelligence Agent.

Its job:

Watch the accounts.

Every morning it looks at:

Then it answers five questions:

That’s it.

You don’t need an agent to “run Customer Success.”

You need an agent to make sure important things don’t get missed.

There are already tools and practitioners experimenting with exactly this kind of workflow. One emerging pattern is using an agent to combine health signals, CRM context, product usage and support history into a single account brief rather than making the CSM hunt across four or five systems.

That’s where I think the near-term value of agents is going to come from.

Here’s a simple version you can try today.

Give the AI whatever account information you have available and use:

You are my Customer Success strategy partner.

Review the account information below (add in all call data, email data, crm data and any other key metrics) and identify the most important changes, risks and opportunities.

Do not simply summarize the information.

Look for patterns.

Specifically identify:

For every recommendation, provide the evidence from the account data that led you to that conclusion.

If the evidence isn’t strong enough to support a conclusion, say so.

—---

The last instruction matters.

One of the biggest mistakes we can make with AI is confusing a confident answer with a correct answer.

The best CS agents shouldn’t just tell you what they think.

They should show you why.

Take 10 customer accounts.

Give the AI access to the same information a CSM normally uses:

Ask it to produce a one-page account brief for each.

Then compare it with what your CSM already knows.

Don’t measure whether the AI sounds smart.

Measure:

How many important signals did it surface that the CSM didn’t already know?

That’s the interesting metric.

Because the real promise of AI in CS isn’t:

“It can summarize my calls.”

We already know that.

The bigger promise is:

“It can see patterns across 100 accounts that no human has the time to see.”

There’s another shift happening that I think is going to be even more interesting.

AI agents are starting to move from preparing the work to executing the work.

For example:

Instead of:

“Here’s a list of customers who might be at risk.”

The future looks more like:

“I found 12 accounts showing risk. I prepared the account briefs, drafted the internal recommendations, identified the appropriate stakeholders, and prepared suggested outreach. Nothing gets sent until you approve it.”

That distinction is important.

Automation isn’t the same as autonomy.

And I don’t think Customer Success should rush toward full autonomy.

Customer-facing communication, commercial decisions, pricing, executive escalations and sensitive relationship moments should still have a human involved.

But there is an enormous amount of work surrounding those moments that doesn’t necessarily require a human.

That’s where agents come in.

Some emerging CS agent platforms are already explicitly building around this model: monitor the account, identify risk, prepare the next action, and keep customer-facing actions approval-gated. 

I don’t think the future of Customer Success is:

Humans vs. AI.

I think it’s:

Humans + agents.

The best CSMs won’t necessarily be the ones who know the most about AI.

They’ll be the ones who understand their customer’s business and know how to deploy AI against the repetitive work surrounding that relationship.

Think about what happens if a CSM who currently manages 30 accounts can effectively manage 60 because agents are:

The CSM doesn’t become less valuable.

The CSM becomes more leveraged.

And that may fundamentally change the economics of Customer Success.

If AI can eventually do 50–70% of the administrative and analytical work surrounding a customer relationship…

What should a CSM actually spend their time doing?

My answer:

Less administration.

Less reporting.

Less CRM cleanup.

Less “just checking in.”

More:

Business strategy.

Executive relationships.

Customer outcomes.

Change management.

Expansion.

Human conversations.

That’s the part of Customer Success that becomes more valuable as AI gets better.

That’s a little around what I’m going to explore here every week.

Not AI news for the sake of AI news.

Not another collection of generic prompts.

I’ll be looking for the things that actually matter to Customer Success:

What changed?

What can you use?

What should you experiment with?

And what does all of this mean for the future of our profession?

If you’re a CSM, CS leader or anyone trying to figure out how AI and agents will change Customer Success, I think we’re going to have a lot to talk about.

See you next week.