I am writing this seeking help/suggestion for problems that we are facing to serve rasa open source assistants for multiple tenants

We are looking for an architectural suggestion to serve rasa bots with our SAAS product where we are trying to offer RASA BOTS as a bot assistant to server the tenant’s users, where a tenant can create its own multiple RASA Assistants dynamically

We are trying to run rasa apps on multiple ports with the help of a reverse proxy (nginx) by mapping the url with the new RASA bot port

The problem with this approach can be everytime a tenant requests for a new rasa bot we have to map the url and restart nginx i think we will face a downtime and will impact other tenant RASA apps as well

Could anyone suggest a solution to server RASA BOTS to multiple tenants on a scale that may have its own set of bots

Thank you