You can just do things: Codex set up Chrome automation scripts to extensively search flights with ITA matrix (~200 possibilities, after choice of outbound/inbound dates & destination airport & booking method)

or, you could use swoop and do it in a second:

bro chose the slowest way to do this

You could always do this before Codex. Also just hit the APIs instead of this UI dog shit.

Oh and this is 1x, not sped up

(this is _not_ using the model's native computer-use)

Why do you need to extensively search flights like this ? Also, Don't you get rate limited or blocked by ITA Matrix after a while ?

Why is it better than something like a Kiwi search?

Can you please give me a pointer where I can learn to do this?

My next codex weekend project:

How? Any chance you’d give a quick bullet point list to an old guy? 😀😅

The underrated part is turning a clever one-off script into a reusable workflow with clear inputs, logging, and failure handling. That’s where “agent magic” starts becoming actual leverage instead of a fun demo.

Three years ago, scraping 200 flight permutations required a Python developer, a proxy network, and a weekend. Today, it requires a 27-second prompt. Expedia and Priceline built hundred-billion-dollar empires by acting as the middlemen for travel data. They aggregated the flights so you didn't have to. Now, you don't need a middleman. You just need an agent. This isn't a lifehack for cheap travel. It is the complete democratization of digital leverage. You used to browse the internet. Now you command a localized botnet to brute-force a corporate pricing algorithm until it bleeds out the cheapest fare. The consumer web spent two decades optimizing interfaces for human clicks. But in 2026, corporations aren't competing for human clicks anymore. They are defending their margins against their own customers' private AI swarms.

Surely you’d have a much better time just reversing the api