👋 Introducing the Rasa Playground

RASA CALM tutorial: issue with integrating API call step of tutorial

You have selected 0 posts.

post by LECARROU on Mar 27, 2025

Hi,

I’m still trying to follow RASA CALM tutorial but the last step when integrating API call to check for sufficient funds before transfer money does not run as expected and transfer pass even if amount pass is greater than balance set in ActionCheckSufficientFunds (1000). Even if not mentioned in tutorial at this step, I tried to re-train rasa before running inspect command. But rasa train lead to error below:

UserWarning: Loading domain from 'domain.yml' failed. Using empty domain. Error: 'Your domain uses an invalid slot mapping of type 'controlled' for slot 'has_sufficient_funds'. Please see https://rasa.com/docs/rasa-pro/nlu-based-assistants/domain#slots for more information.'
... 

post by m_ashurkina on Mar 27, 2025

Hi @LECARROU, thanks for your message. Could you please show how your flow is structured? Do you use the action in the flow action_check_sufficient_funds?

post by LECARROU on Mar 27, 2025

I go ahead trying to implement “basic_flow_with_branching” and when I train model, I notice in logs that “action_check_sufficient_funds” has been registered and it is now working. I am a quite disappointed not to understand why sometime it works and other not

post by m_ashurkina on Mar 28, 2025

Hi @LECARROU thank you for your response. Just checking whether you’ve seen this page Rasa Pro Tutorial Rasa Documentation? It’s a handy tutorial on how to build the CALM assistant from scratch.

post by LECARROU on Mar 28, 2025

Hi Marina,

Yes, I did the tutorial and it works “issues” came when I tried to start a new calm project (rasa init --template calm) which config seems to be quite different. I came with issue with OPENAI quotas restrictions… I try to bypass this issue implementing Ollama LLM models but until now I still have issues with OPENAI KEY even if normally config do not refer to OPENAI.

config.yml

recipe: default.v1
language: en
pipeline:
  - name: CompactLLMCommandGenerator
    llm:
      model_group: ollama-gemma3-1b
policies:
  - name: FlowPolicy
  - name: IntentlessPolicy
  assistant_id: 20250328-141853-unary-vial

endpoints.yml

action_endpoint:
  actions_module: "actions"
model_groups:
  - id: ollama-gemma3-1b
    models:
      - provider: ollama
        base_url: http://localhost:11434/
        model: gemma3:1b
        request_timeout: 7
        max_tokens: 512

post by m_ashurkina on Apr 7, 2025

Hi @LECARROU thanks again for your question. If you want to use text-embedding-ada-002 for your Enterprise Search you’d need to add your OpenAI API Key, or you could try using another embedding model which is available from Ollama. We’ve recently published a tutorial on how to customise Enterprise Search parameters.