Model Practical ML Usecases As Model Chains in Autotune

[KDwevedi]

Description

For a given recipe/flowise flow/bot logic def in Bhasai, one or many ML models may be used to generate a single response.
In case of multiple models being used for a bot recipe, we want to be able to visualise these interdependencies in admin panel. Therefore, there is a need to be able to represent these interdependencies as config within autotune.

Usecases

• Being able to visualise model input and output
  • [ASR] (iterate) -> [Transliteration] -> [Translation] -> [Intent Classification] -> [NER] -> [T2S]

Dependencies

• AI Tools Proxy should be used for all AI-Tools calls and model input-outputs be tracked independently of frontend flows using a telemetry integration.
  • https://github.com/BharatSahAIyak/ai-tools-proxy/issues/13