Hi Nandhini,
There are various factors related to the overall architecture, configuration, or integration between services.
Use Azure Application Insights or Power Platform’s inbuilt telemetry to log message processing times at different stages.
This will help pinpoint the stage where the delay is occurring.
You can reduce the number of topics or intents for faster recognition, or switch to a pre-built, optimized model.
Check if your Power Platform or Azure resources are sufficiently scaled to handle the load. If you notice the system slows down at peak times, you might need to increase the available capacity or add redundancy.
If your bot connects to external systems (e.g., Dataverse, SharePoint), network latency might be contributing to the delay. Ensuring that your bot has quick and efficient access to these systems could help reduce the delay.
Some bots have an "initial warm-up" period, especially after being idle for a while. This could cause the first interaction in a session to be slower than subsequent messages.
Thanks.
