Genesys Cloud - Developer Community!

Hello we are leveraging Genesys Bot Connector to integrate with our internal AI Agent.

While running through some tests we are seeing that postUtterence responses have around 1.5 seconds of latency from when the message is sent to when Genesys shows the response in the Messenger UI versus a latency of .25 seconds through local curl. 

What is the underlying cause of this latency? Are there any configurations we can leverage to shorten this time? 

Hey @Meg Luong,

Great question - we've actually run into similar latency differences in our testing as well. Based on what I've seen in the docs and community discussions, that ~1.25 second delta you're experiencing is pretty typical and comes from a few different layers in the Genesys stack.

The main culprit is the processing chain itself. When you hit your AI agent directly with curl, you're just measuring your bot's response time. But when it goes through Genesys Bot Connector, the message has to make a few extra stops:

- The utterance comes from the Messenger UI over WebSocket.
- Genesys processes it through their messaging flow in Architect.
- Bot Connector then relays it to your external bot via the postUtterance webhook.
- Your bot responds back through Bot Connector.
- Bot Connector translates the response format.
- It goes back through Architect.
- Finally gets rendered in the Messenger UI.

Each of these hops adds some overhead - serialization/deserialization, protocol translation, routing, etc. The 10-second timeout mentioned in their docs gives you an idea of what they consider acceptable, so your 1.5 seconds total is well within normal range.

One thing you might want to check - the default postUtterance timeout is 11 seconds, but they can extend it up to 60 seconds if needed. However, that's more about preventing timeouts than reducing latency.

For actual latency improvements, there isn't much configuration-wise you can tweak on the Genesys side unfortunately. The main optimizations would be on your end:

- Make sure your bot endpoint is in the same region as your Genesys deployment (check their IP ranges for your region).
- Optimize your bot's response time even further if possible.
- Consider caching common responses.
- Ensure your infrastructure can handle the webhook load without queuing.

If the 1.5 seconds is causing real UX issues, you might want to look into their Messenger SDK to add typing indicators or other feedback to make the wait feel shorter to end users.

Hope this helps! Let me know if you want to dig deeper into any specific part of the flow.