Vertex AI Search adds new productive AI capabilities and enterprise-ready features to meet one-stop search needs.
With Vertex AI Search, Google Cloud’s out-of-the-box search solution, you can solve your need for a search application on your enterprise data in just a few days or even a few hours.
The tunable Retrieval Augmented Generation (RAG) system provided by Vertex AI Search will lead you to information with the ability to add custom embeddings and harness the power of large language models.
Custom search to suit business requirements
Customizable answers: With customizable answers, which are currently in preview, you can control how the results are generated and have the answers returned in different formats such as “Short”, “Detailed”, “Casual” or “Formal”.
Search tuning: A preview of which document will come first in enterprise document search will be available later this month. With a training set of 50 – 100 questions/answers, you can enable Vertex AI Search to provide a more precise ranking and better answers.
DIY search engines with Vector Search and Vertex AI Placement : If you have more complex and customized use cases, you can use Vector Search and Vertex AI Placement to create your own search, recommendation and other productive AI applications.
- Vector Search: Indexes the data as it is placed, providing very fast access to the results associated with the search. The user interface has now been updated to help you build your own vector search system. In addition, the time for small number indexing has been reduced from hours to minutes and filtering features have been improved.
If you want to get more detailed information, you can contact us. - Vertex AI Placement: Video support for multi-model placements that support text and image placement is now available in preview. By having all three input types sharing the same semantic space, you can create new use cases for video files.
Linking results to data: new options for accurate results
The possibility of hallucinating productive AI is a concern for many businesses. To counter this, Vertex AI Search offers several options for grounding results in data.
- Grounding with business data: Generating results using your business data and supporting them with summaries and citations will help your users to check the accuracy of the results. This feature is at the core of Vertex AI Search and can be enriched with other data sources that you can connect using Vertex AI Connectors.
- Baselining with open source data: In a business that usually accesses open source data such as Wikipedia, baselining tests are ongoing so that the data can be included in search results. Thus, you can save time and effort to look in more than one place by being able to search not only business-specific but also general data from a single place.
Looker Studio Pro is now available on Android and iOS
You have an idea in the middle of your coffee break and you want to back it up by reviewing your reports created in Looker Studio Pro. No need to go all the way to your desk, you can access your reports conveniently from your mobile phone using the mobile-friendly Looker Studio app.
The Looker Studio app, which you can download from Google Play and the App Store, will provide you with enhanced ease of use and improved reading comfort.
Google Cloud demonstrates the power of TPU infrastructure by showcasing the world’s largest distributed learning job for large language models
Along with generative AI, the underlying large language models have grown very rapidly, with billions of parameters and trillions of training particles. Along with this has come the increased use of AI acceleration chips, which underlie the supercomputing power required to train such large language models, often distributed across large clusters. Managing such large-scale distributed machine learning has brought many common and important technical challenges.
In response to these challenges, Google Cloud has made “Cloud TPU Multisectional Training” available to everyone.
Click here for details.
You can enhance the search and machine learning experience with the newly announced text analytics and preprocessing features in BigQuery.
BigQuery can enable you to perform valuable business analytics using Search and Machine Learning on your massive amounts of textual corporate data.
Text preprocessing is a critical step in extracting meaningful and valuable information from raw text. The quality of the output produced by text preprocessing required for text-based operations such as search and machine learning streams depends on factors such as how the operation is performed and which algorithm is used.
In addition to text processing, which is a core part of BigQuery at this crucial stage, you can also use the new text analysis and preprocessing functions that are currently in preview.
Click here for details.
