Could an AI assistant really generate FortiGate setup scripts?

[cbum63]

I’ve been working on FortiGate CNF deployments on AWS lately and honestly, writing all the config scripts by hand is starting to feel like I’m doing more copy-paste than actual engineering. Out of curiosity I tried using an AI tool (deepseekai.free) to see if it could generate some drafts for me. To my surprise, it actually spat out a decent starting config – and even slipped in a joke about YAML being harder than coffee, which made me laugh while still being useful.

It made me wonder: has anyone here experimented with AI assistants to generate FortiGate setup scripts? Could be for Firewall Manager, CloudFormation, or even policy deployments. I can see the appeal of having a rough draft done in seconds, but at the same time, trusting AI-generated configs feels like asking your intern to run production changes.

Curious if others have tried this. Did it save you time, or did you end up rewriting everything anyway?

[jerrysa]

---

@cbum63bpl schedule wrote:

I’ve been working on FortiGate CNF deployments on AWS lately and honestly, writing all the config scripts by hand is starting to feel like I’m doing more copy-paste than actual engineering. Out of curiosity I tried using an AI tool (deepseekai.free) to see if it could generate some drafts for me. To my surprise, it actually spat out a decent starting config – and even slipped in a joke about YAML being harder than coffee, which made me laugh while still being useful.

It made me wonder: has anyone here experimented with AI assistants to generate FortiGate setup scripts? Could be for Firewall Manager, CloudFormation, or even policy deployments. I can see the appeal of having a rough draft done in seconds, but at the same time, trusting AI-generated configs feels like asking your intern to run production changes.

Curious if others have tried this. Did it save you time, or did you end up rewriting everything anyway?

---

Yeah, AI tools can definitely speed things up for FortiGate setups they’re great for drafting baseline configs or templates, especially for repetitive tasks.

But you’re right, they still need human review before deployment. I usually treat AI output as a starting point, then fine-tune it for security and environment specifics.