Is AI Killing Trust in SaaS?

Why this matters

Someone reached out to the show asking to be a guest. The message was thoughtful, personalised and referenced a specific episode. When they got on the call, they admitted they had never listened to the podcast — AI had written the whole thing. That story is not an outlier. It is becoming the default operating model for a lot of SaaS teams, and it creates a problem that starts in outreach and runs all the way through to how CS teams understand their customers.

Key ideas

AI scales bad behaviour, it does not change it. False familiarity, inflated claims and manufactured credibility have always existed in sales. What AI does is make them faster, more polished and more widespread. The outreach looks great. The follow-up conversation exposes that there is nothing behind it. You get the meeting but you lose the trust.

Think of AI like a college intern. If an intern representing you sent a message full of things that were not true, you would have a coaching conversation. You would not give them less oversight — you would give them more. AI needs the same. Someone has to review what goes out, own the consequences and use judgement to correct what AI gets wrong.

Polished is not the same as accurate. In customer success, AI can produce a QBR deck that looks executive-ready. The account narrative sounds coherent. The health score looks plausible. But if the interpretation is wrong — if usage has dropped because a team consolidated and not because the product is failing — that polished story is going to damage the relationship, not protect it. Customers do not judge you on whether you used AI. They judge you on whether you understand them.

Data shows activity. It does not show intent. Usage declining, support tickets spiking, health scores going amber — none of these data points tell you what is actually happening in the customer's organisation. The budget holder may have changed. There may be a new executive proving they are tough on spend. There may be a reorganisation no one mentioned. AI cannot know the politics. The CSM has to.

The guardrails most teams are missing. Every AI-generated output going to a customer should have a named human owner. Facts and interpretation should be separated — AI can flag a pattern, but a human needs to validate what it means. And CSMs need to be trained not just to write better AI prompts, but to challenge AI output. The dangerous answers will not look wrong. They will look very polished.

"AI is not there to help you pretend you understand your customers better. It's there to help you understand them better." Those are very different things. One creates scale theatre. The other creates actual capability.

The Breakthrough Challenge

Where in your business are you using AI to create the appearance of preparation, personalisation or expertise — and what human verification step do you need to add before it reaches a customer, a prospect or an employee?

Questions for SaaS leaders

• If your AI-generated customer outputs were wrong, how quickly would you know — and who owns the accountability?
• Are your CSMs being trained to challenge AI output, or just to produce more of it?
• Where in your customer journey could polished AI communication be masking a real understanding problem?