I am currently building a Proof of Concept in Genesys Cloud. I am incredibly frustrated with the Architect tool. I am trying to build a dynamic menu where the flow queries a REST API, receives a JSON list of available product names, and then uses a “Play Audio” action to speak those names to the caller using Text-to-Speech (TTS). However, if the JSON response contains an array with more than three items, the Architect flow instantly errors out and drops the call with a “System Error”. I have verified the JSON parsing is correct. Why does Architect crash when attempting to iterate and play a dynamic list of strings?
I have encountered this precise limitation. The issue is not your JSON parsing; it is a limitation of the Architect expression engine. You cannot simply pass a large, dynamic array of strings directly into a single “Play Audio” node using TTS. The expression parser has a maximum evaluation limit. To solve this, you must use a “Loop” action. You must determine the length of your JSON array, initialize a counter, and use the loop to play one item at a time (e.g., MakeAudio(Task.ProductList[Task.Counter])), incrementing the counter on each pass.
I develop API integrations and I can confirm The point above is correct. You have to break it down. I will add one more technical detail: if the total combined string length of all your TTS audio exceeds 2048 characters, the TTS engine itself will reject the request and throw a system error. If your API is returning long product descriptions along with the names, you will hit that limit very quickly. Looping through the array item by item is the only safe way to ensure the TTS engine does not choke on the payload size.