I asked 4 AI models which observability tool to use. They named Datadog every time and skipped the modern ones.

I build a tool (Bersyn) that measures which products AI models recommend in a category, so I run a lot of these scans. This week's was observability, and it was the cleanest example yet of a pattern I keep hitting.

I asked ChatGPT, Claude, Gemini and Perplexity the question a real engineer types: what should I use to monitor and debug my SaaS in production. Then I counted who actually got named, each tool measured in its own home category.

Better Stack: named in 0 of the answers, on all four models.
Axiom and Highlight: only Claude, nobody else.
OpenStatus: only Perplexity.

What gets named instead, every time: Datadog, Splunk, Pingdom, LogRocket, Cachet. The names that owned the monitoring conversation around 2016.

The part that makes it a real problem rather than a funny one: AI is not clueless about the category. It names Sentry in 100% of ChatGPT answers and knows Honeycomb across all four models. So there is room in its map for modern tools. It has just frozen that map around the old incumbents, and everything newer is invisible.

For indie founders this is the scary version of the SEO problem. Search took twenty years to learn you exist. AI is forming its recommendations right now, off thin evidence, and it defaults to whoever dominated the content a few years ago.

Two genuine questions for this crowd:

Has anyone building a dev tool actually seen a buyer arrive after asking an AI for a recommendation yet, or is it still too early to matter?
If you have tried to change what the models say about you, what actually moved it?