Sure, but there's a lot of other things you're looking over:

- We support selenium with the same version of Chrome as puppeteer's. Everything is versioned together.

- Queueing/Concurrency and notifications. Arguably done in other efforts, but works.

- Numerous other APIs built on top of puppeteer to do core cases. No need to write your own integration.

- Monitoring and more.

People do pay for this in order to not think about the burden of supporting it. Managed databases are a thing, even though they are freely available to download and run.

I'm a browserless user, and the headache for me was cost. You're right that it's dirt-simple to spin up a puppeteer service in docker, and that's what I had been doing previously. But, for my usage, I found it was cheaper to pay browserless than to run my own EC2 instances. Granted, I probably account for a very, very tiny fraction of that $24k/mo :)

I dove into the comments to ask the same question. I just did something similar to this with a few lines in my docker-compose.yml, which eventually turned into a few lines in a Helm chart. Why would I pay someone money for something as trivial as this?

Now, I won’t write another rant on the subject of Eternal September, and how our field is increasingly populated by charlatans who can’t do simple things on their own. But yes: that’s exactly how I feel.

Disclaimer: Neither I nor my colleagues or anyone I'm aware of are, were, or ever will, use said service.

Did you divide (time that took you to complete this trivial task + time to document it, version it, explain it to coworkers, maintain it next time chrome decides to break something) * (your $/h cost) / 0.00008 before doing that?

Here's a build vs buy calculator to save you from doing the maths: https://baremetrics.com/startup-calculator

This is a dangerous calculator. It misses the cost of deployment, integration, internal support, etc.

(This can be significantly larger for the purchased solution then an internal one as well)

Yep. Puppeteer is the state of the art wrt headless browsers. There's money to be made selling B2B wrappers of OS software though. Not to trivialize the author's work...

People who would rather pay $30/mo to have someone manage it than run it themselves.

That's a good deal if it takes more than about 10 minutes of my time a month to run it myself. Wanna venture a guess on how many years to ROI on time put into initial setup?

And fwiw I do run my own puppeteer cluster because it's economically advantageous at our scale.

Yea, I’m thinking making about a company named after some kind of low hanging fruit and just building out a bunch of these trivial little use cases that keep popping up and put them all behind a subscription model.

You’ll probably see me with an article in 6 months about how I’m making $150k/mo for about 2% of my time.

I have set a calendar item for 6 months from now. Looking forward to your blog post.

Feeling the pressure now.

Or you're just gonna leave a snarky comment, and others gonna build those services and get the money...