A startup is building a customer support chatbot using Vertex AI and wants to ground responses in their product documentation to reduce hallucinations. Which approach should they use?
Trap 1: Use the Codey API for text generation.
Codey is for code generation, not grounding documentation.
Trap 2: Use the base model without any grounding to maximize flexibility.
Lack of grounding increases hallucination risk.
Trap 3: Fine-tune the model on the documentation and deploy.
Fine-tuning alone does not guarantee real-time grounding.
- A
Enable Vertex AI Grounding with a custom enterprise data store containing the documentation.
Grounding ties responses to specific documents, reducing hallucinations.
- B
Use the Codey API for text generation.
Why wrong: Codey is for code generation, not grounding documentation.
- C
Use the base model without any grounding to maximize flexibility.
Why wrong: Lack of grounding increases hallucination risk.
- D
Fine-tune the model on the documentation and deploy.
Why wrong: Fine-tuning alone does not guarantee real-time grounding.