Context: Leveraging enterprise knowledge in AI prompts is essential for generating accurate and relevant responses. However, integrating this knowledge into generative AI models can be challenging as it is embedded in Enterprise Databases. Custom prompts enable makers to use generative AI models addressing various types of content generation scenarios. These models use their default knowledge included in their training data to answer. However, this knowledge isn't sufficient to deal with use cases requiring business specific data context.

Solution: Using Dataverse to add enterprise knowledge to generative AI prompts ensures that the AI has access to comprehensive and up-to-date information. Dataverse provides a centralized repository for storing and managing enterprise data, which can be seamlessly integrated into AI models. With this capability, makers can add Dataverse data records as an input source to their AI Builder GPT prompts. This allows users to customize the knowledge from GPT with enterprise data stored in Dataverse. In prompt builder, an option is available to add knowledge to GPT prompts. For example, you can select specific Dataverse records to include in the prompt, along with specific instructions to filter the data according to the target scenario. This enables you to amplify GPT with data knowledge. You can also perform key scenarios like data summarization and classification through prompts that can be triggered from copilots, Power Automate, and Power Apps.

Impact: Here Data Retrieval Augmented Generation (RAG) allows enterprises to provide external information to augment the knowledge of the model. This augmentation can result in getting the answers. This integration enhances the accuracy and relevance of AI-generated responses, leading to better decision-making and more efficient operations. It also ensures that the AI can provide insights that are aligned with the organization's knowledge base, improving overall productivity and effectiveness. The number of scenarios enabled by this capability is limited only by  creativity! 

The following list provides some examples.

1. Create a summary of the account named Name using only these columns: Account.Name, Account.Description, Account.Orders (Order).Name, Account.Orders (Order).Amount.

2. Classify the Email into one of these Category.Name matching based on Category.Description.

3. Draft a reply to this Problem matching data from FAQ.Topic and getting inspiration from FAQ.Solution.