UPDATE: It’s been running for several days now and it has generated over 300 blog posts. So far the costs have been negligible (less than 1 USD). I’ll keep posting as AWS bills me.

amaynez/Make.com-to-AWS: This is an experiment to try to recreate a complex Make.Com scenario using AWS Step Functions (github.com)

Original Post:

I have a business and this business has a blog, I got the idea of using LLMs to write SEO relevant blog posts. I discovered Make.com and in no time I was fully automated. I purchased the 10,000 operations a month plan, and it always runs out by day 20 of the month, because I also have other automations like creating LinkedIn posts on my personal page, doing some sync on some database backups between airtable, notion and Google sheets, and some deduplications. It is a fantastic tool, and I love it. Specially because I can do quite complex things with it very quickly. I now want to increase the frequency of the blog posts scenario but that would mean that I have to purchase a larger plan, even after optimizing all my scenarios for minimal operations (for instance the blog post one originally used up 230 operations per run, and now it only uses 57). So I did the math and it turns out that using AWS Step Functions and some Lambdas I can have the same functionality but much cheaper (if you compare the spend per operation). Obviously at the great cost of taking much longer to program everything. For instance parsing JSON which is just a piece of cake in Make.com, is a pain when coding it yourself. Im between jobs at the moment so I am taking this opportunity to learn how Step Functions work in AWS by recreating my Make.com scenarios. I will then run them for a month and see if it is truly cheaper or not. But in the mean time at least I spent some time sharpening my python skills and learning some cloud development which is completely new to me. I have spent about 4 hours and haven’t even finished the blog post scenario. When I finish it I will post the GitHub repo with a reference to the Make.com scenario it replaces if anyone is interested. I think that for use cases like mine where I don’t need to change or create scenarios constantly, but have a more stable automation landscape, it make sense to go from no-code to code. Otherwise, make.com is brilliant to create scenarios quite quickly. Let me know if it makes sense to you.

EDIT:

This is my Github Repo: amaynez/Make.com-to-AWS: This is an experiment to try to recreate a complex Make.Com scenario using AWS Step Functions (github.com)

The scenario is up and running.