It is the difference between setting up your own server to communicate with the CPaaS vendor and run all comms logic on your service versus having it reside and oeprate on the CPaaS’ vendor’s machines.

The same reasons for picking AWS Lambda versus putting a Node server on an AWS EC2 machine will apply here with the nuances of this being a communication scenario.

P.S. I tried to reply in thread but was labelled as spam.

Aivis,

I think you are missing a lot of the point here. One of the ways to invoke these serverless functions or drag and drop interactions isn’t by way of an API call – they can also take place through an event on your “numbers” – an incoming call or a message for example. As such, they make an awful lot of sense to use that way, especially for the smaller types of interactions or for those that require quite a lot of scaling to implement properly.

At testRTC we are using a flow tool for some of our customers who need to manage incoming calls in their tests and monitoring scripts. I’ve also seen these tools used by other vendors in production for daily operations. In some cases, they happen with no API interaction. In others, they do occur alongside API calls where their purpose is to reduce the development effort and ease maintenance.

Jaro, for the time being, AI/ML is slow moving in CPaaS. Many talk about it but do rather little, or less then you’d think they would with the data they are sitting on.