Home » ai

Tag: ai

Building ZAPs!

Since I recommended you build an API into your service that integrates with Zapier, I figured I might as well give it a try myself, and not just talk theory

I was shocked to finish the integration without having to write one single line of code

But I am getting ahead of myself. 

Part of my skill up plan is to take a deep dive into automation. What a better way to study than looking at what the masters are doing? Or, in this case: Zapier. 

I’ve fired up their tutorial and was getting ready to crank some code! 

I have integrated APIs before, and if the documentation was clear and with relevant examples, I could get the hang of it pretty quickly and then extend it easily. 

As I am progressing through the tutorial, it dawns on me that this integration will be created entirely online, in the Zapier UI, by answering some questions and filling forms! 

My mind was going: “it cannot be that simple! Each API provider has its own interface or authentication mechanism… you would have to write some adapter for that!” 

No, sir! Not so! 

If the APIs follow the standards when it comes to authentication and passing data around using JSON, then Zapier has you covered! 

You answer the questions, you fill out the forms as instructed… and you’re done! No code required! And since there is no code required, there is nearly zero chance of having a bug in your integration. How cool is that? 

Needless to say, I am impressed. And I got another confirmation of why following standards is a good idea. 

Not quite like building with Legos

Even though I did not write any code, it did not feel like building with Lego blocks. But more like solving a jigsaw puzzle. Because the APIs are very diverse, you need to spend some time to connect the right pieces in the correct order. But since everything is built via filling forms, you have validation available and tests. There is immediate feedback if what you are making is not working as you’d expect.

I did spend two hours getting the final step to work. I almost contacted Zapier’s support to inform them that their builder is broken, but in fact, the problem was with me :). 

I had a typo in one of the data fields, and the error message I was getting from the API I was using at the time was very misleading (apparently on purpose). So it took me two hours to find that typo in my configuration and get the Zap to work. Take away here: please make your error messages clear; it will save you time in the future!

Just as with finishing a complex puzzle, seeing the success message was so satisfying. 

Why all the fuss?

Building a Zapier integration is so simple and straight forward that I can easily see AI building integrations and ZAPs in the (very) near future. This means that if you decide to provide an API that can integrate with platforms like Zapier, your service may be used in ways you cannot even imagine. The pieces can be interconnected in infinite ways to create something new, exciting, and why not: disruptive!

Keep on creating the new paradigm.