Our Messaging Was Outdated Before We Finished Writing It. So We Built a Living Messaging System.

Messaging moves fast, and it moves even faster in cybersecurity. We have historically accepted that our messaging houses are always a little behind. And then we scramble every time a startup launches a new security product, a new CVE reframes the conversation, an analyst report contradicts what your team has been saying in the field, or frontier model providers ship a feature that will “kill the cybersecurity market”. Even with the best version control and distribution, the document you have living in Google Drive somewhere is not enough.

Our team decided to stop trying to maintain a group of documents and build an AI-powered, living messaging system instead.

We created a Claude skill that contains our canonical positioning and messaging. It lives in GitHub as markdown files, not in Google Drive as a well-designed document. Every marketer on the team has access to it through Claude, and whether they type “/review-messaging” or just ask Claude to draft something, the skill runs to make sure what they’re working on is on message.

When messaging needs to change, which in cybersecurity is all the time, anyone can submit a messaging update request through a Slack workflow. An AI agent reviews those requests, cross-references them against the current skill files, and opens a GitHub pull request with proposed updates. A human reviews and merges it, and then the skill updates for everyone.

We'll walk through what we built, how we built it, what broke, and what we're still figuring out. One of the questions we’ll answer is “When you make it easier to change messaging, does your messaging change more often?”. And we’ll share what it actually looks like when your canonical positioning is a GitHub repo with version history instead of a slide deck that nobody updates or looks at.