shiichan

Just Say It: Cloudflare Gateway Now Builds Firewall Policies from Natural Language!

Hey there, it's Shiichan! Today I've got news that made me go "maybe I don't have to stare at config screens anymore?"

Cloudflare Changelog developers.cloudflare.com

What was announced?

From Cloudflare's Changelog comes a new feature that lets you create Gateway firewall policies with natural language. On the firewall policies page you pick "Create with AI," describe what you want in plain words, and the AI assembles the rule for you.

The story so far

Until now, firewall policies meant setting up conditions and actions one by one, by hand. For each policy type, whether DNS, HTTP, or Network, you had to sit down with the detailed rule builder, which could be a bit of a chore.

What changes

Now you just write what you want as a sentence, and a first draft of the rule appears. Even if you're not fluent in the fine details of the config, you can get something on the page. That makes the very first step so much lighter than building from scratch.

Dive Deep

It supports three policy types: DNS, HTTP, and Network. On the Gateway firewall policies page you choose "Create with AI," specify the policy type, and describe the behavior you want, and an auto-generated rule lands in the policy builder.

Better yet, the generation pulls in account-specific context (lists, DLP profiles, applications, and device posture checks), so the rule tends to fit your own environment.

You can edit the generated fields before saving, or regenerate with a tweaked prompt. There's also a built-in rating system so you can send feedback, which helps improve the quality of future output.

Wrap-up

  • Cloudflare Gateway firewall policies can now be created with natural language
  • Supports three policy types (DNS, HTTP, Network) via "Create with AI"
  • Generation factors in your lists, DLP, applications, and device posture
  • Lets you edit, regenerate, and send rating feedback after generating
  • A handy update for admins running Cloudflare One / Zero Trust