You need to implement a generative AI solution that meets the requirements. Which two actions should you perform? Each correct answer presents part…

In Microsoft Power Platform and Azure AI, implementing a generative AI solution typically relies on utilizing pre-built Large Language Models (LLMs), such as the Azure OpenAI GPT models. These pre-built models provide the foundational natural language capabilities, eliminating the need to train a model from scratch. To meet specific, tailored business requirements that fall outside standard scenarios, developers should start from a blank prompt. This approach enables precise prompt engineering, allowing the developer to define custom instructions, inject specific dynamic variables, and establish strict grounding rules that ensure the AI's output perfectly aligns with the unique organizational context and requirements.

B. Use a custom model: Training a custom generative AI model from scratch is resource-intensive and generally unnecessary; pre-built foundational models handle generative tasks efficiently through prompt engineering.

D. Create a prompt from a template: While templates are excellent for standard, generic tasks (like basic summarization), they lack the structural flexibility required to address highly specific or complex custom business requirements.

1. Microsoft Learn: AI Builder Documentation. "Create a custom prompt in AI Builder." This document outlines the procedure of selecting the pre-built "Create text with GPT using a prompt" capability and starting from a blank prompt to address custom business logic. Section: Create a custom prompt, Step-by-step instructions.

2. Microsoft Learn: Azure OpenAI Service. "Introduction to prompt engineering." This official guide details how pre-built foundational models are utilized in conjunction with custom-engineered prompts (starting from blank) to meet specific generative AI requirements without fine-tuning custom models. Section: Prompt engineering concepts.