20260425 Mom

MOM

• ai components in mule

• a -> b; b -> a; x -> y

• psafe

• better than using models because we have proven migration results

• with normal AI X-tokens; with us Y-tokens

• 600 apis in hand

  • apigee2mulesoft
  • tibco2mulesoft

• process

  • feed to apigee and mulesoft documentation - via RAG
  • plan, migrate and validate

• apigee and apigee edge as xmls. refer policies. kv map - object store. proper prompt to llm.

• policy mapping layer and routing for lightweight and heavyweight models

• mulesoft project templates

• 4-5 policy mapping and migrate sample apis

• go to 20-30 apis via tuning

mapping layer policy RAGs for all platforms model routing layer migration planning layer migration execution validation layer

Needed

• doc for techstack
  • core: pydantic-ai
  • llm provider: openrounter
  • vector database: pinecone
  • validation framework: pytest and pydantic
• mulesoft documentation
• apigee documentation