Proof, Not Hype

AI in customer success, reported with the receipts

I am Colin Slade. I am SVP of AI Strategy and Customer Success at Cloudbeds, a hospitality technology company, where I co-lead a customer success organization of about 250 people. I write up what we build with AI, what it cost, and the parts that did not work.

Who this is for

Post-sales leaders in business software who are being told to adopt AI and have not been handed a playbook. Customer success, support and customer experience operations, onboarding, enablement. If you are sitting on tickets, transcripts or churn data you have never had time to mine, that is the starting point I keep coming back to.

Nothing here assumes you are behind. I feel behind most mornings. The useful question is not how far ahead someone else is, it is which pile of data you already own.

The strongest evidence I have

Over the past year my team at Cloudbeds built more than 150 internal AI agents, automated roughly 7,000 hours of work each month, and delivered millions in efficiency gains without adding headcount. Two of those builds are written up in full, with the costs:

• Workflow study

  Reading every churned account at once, for about $100

  A churn analysis across millions of data points, run in roughly an hour for roughly one hundred dollars, which found that a large share of recorded churn was not product churn at all.

• Workflow study

  Triaging 650 support tickets in 30 minutes

  An AI triage run across 650 Zendesk tickets that categorized and sentiment-scored every one, cross-referenced four other systems, and found that between 30 and 40 percent of the tickets did not need to be open.

• Podcast · [Un]Churned · Gainsight

  How a 5-Person Team Built 150 AI Workflows That Changed an Entire Company

  A long conversation about how the agents actually got built, and by how few people.

• Framework · version 0.1

  Patterns, Not Tickets

  A method for reading operational data a team already owns at full scale instead of one row at a time, so the output is a system fix rather than a queue of individual fixes.

The short version of all of it: when you look at tickets one by one, you fix issues. When you look at them together, you fix systems.

One CTA, and this is it

Field notes on what is actually working with AI in customer experience, including the costs and the runs that failed. Roughly weekly. No pitch at the end of it.

Your email Subscribe

Form endpoint not yet wired The signup form is built and renders without JavaScript, but it has no action URL yet. Supply the HighLevel form endpoint and set NEWSLETTER.confirmed to true in src/consts.ts. Until then the field is disabled on purpose, so no subscriber is collected and lost. In the meantime, follow Colin on LinkedIn.