Google has announced two collaborations to guarantee more accurate and dependable data for AI apps and agents. Google is collaborating with Moody's, MSCI, Thomson Reuters, and Zoominfo, according a blog post from the business.

For improved AI answers and outcomes, Google's Vertex AI will leverage third-party data from Thomson Reuters and Moody's. This data will be available for developers to use in order to guarantee that responses are given with greater accuracy and fewer "hallucinations."

According to Google, Vertex AI will launch a new service that will enable users to root their models and AI agents using specialized third-party data. Google hopes to reassure businesses that their AI agents will be connected with data from third parties, such as Thomson Reuters and Moody's, as some are still hesitant to use them.

Additional new AI capabilities for businesses

Additionally, Google unveiled a number of new tools that it refers to as "grounding capabilities" to assist enterprise clients in creating agents and apps that are more powerful. One new feature that will help customers balance quality and cost-efficiency is Grounding with Google Search. It will intelligently choose when to utilize the model's training data and when to use Google Search results.

Additionally, there is grounding with high-fidelity mode, which generates answers by focusing on context provided by the user using a Gemini 1.5 Flash model that has been fine-tuned. Important enterprise use cases that the service enables include data extraction from a corpus of financial data and summarization across numerous documents. Higher degrees of factuality and a decrease in hallucinations follow from this.

Sentences in the answer that support the given assertions have sources associated when the high-fidelity mode is used. Additionally, grounding confidence scores are given," Gokturk wrote in a blog entry.