QuestionQ84

Assembling and Deploying Applications

A Generative AI Engineer is helping a cinema expand its website chatbot so it can answer questions about specific showtimes for movies currently playing at the user’s local theater. Location services already provide the user’s location to the agent, and a Delta table is continuously updated with the latest location-specific showtime information. They want to add this capability to their RAG application.

Which option accomplishes this with the least effort and the best performance?

  • A Create a Feature Serving Endpoint from a FeatureSpec that references an online store synced from the Delta table. Query the Feature Serving Endpoint as part of the agent logic / tool implementation.
  • B Query the Delta table directly via a SQL query constructed from the user’s input using a text-to-SQL LLM in the agent logic / tool implementation.
  • C Set up a task in Databricks Workflows to write the information in the Delta table periodically to an external database such as MySQL and query the information from there as part of the agent logic / tool implementation.
  • D Write the Delta table contents to a text column, then embed those texts using an embedding model and store these in the vector index. Look up the information based on the embedding as part of the agent logic / tool implementation.
Explanation

Structured RAG applications can use an online table hosted through a feature serving endpoint, with an agent tool performing low-latency lookups. Synchronizing the continually updated Delta-table data to an online store avoids external replication and the overhead and uncertainty of LLM-generated SQL, while preserving exact, current location-based showtime retrieval rather than relying on semantic vector similarity.

Learn more

Community Discussion

No comments yet. Be the first to start the discussion!