In the last year, the community has grown by almost 70,000 people. That number is driven overwhelmingly by the influx of folk from domains like observability, security, AI and ofcourse… data.

For all the progress we’ve made in the platform engineering industry, there are some bottlenecks that slow teams down to a crawl and receive almost zero awareness. One big topic is access to usable, realistic, safe test data. Even the most advanced Internal Developer Platforms grind to a halt when engineers can’t replicate production behaviours, or test any features without falling apart.

While everyone was having fun at re:Invent, I was diving into Mark Brocato’s awesome article and webinar combo on the value of synthetic data, and how to effectively generate it. The key points are:

• Many teams have long relied on masked production data or rules-based synthetic datasets but these are usually too rigid to capture the messiness of real systems.
• We’re now entering a new phase: Agentic AI that can reason over schemas, write code, enforce relationships, and iteratively generate far more realistic test data.
• This change is crucial as platforms scale with microservices, distributed states, and AI-driven features, the gap between “having an environment” and “having the right data in it” has never been more important.

The result is a subtle but important shift in mindset, as test data becomes an integrated capability of the platform, not an afterthought. As agentic AI matures, the platform foundations around it grow stronger, the teams who embrace AI effectively will find themselves able to move faster, test more thoroughly, and maybe provide some of that sweet sweet self service we all keep talking about.

Now wouldn’t that be something.