Google has made an important announcement regarding its generative AI support in Vertex AI, the company’s robust machine learning platform. This development signifies a significant milestone as these AI capabilities are now generally available to developers. With models such as PaLM 2, Imagen, and Codey at the core, Vertex AI provides a host of powerful features that enable developers to generate and classify text, create multi-turn chat experiences akin to ChatGPT, and leverage a text embedding API for tasks like semantic search and recommendation engines.
To make these capabilities easily accessible, Google has integrated these models into the Vertex AI “Model Garden,” which serves as a repository of enterprise-ready foundational and task-specific models. Additionally, the models are available within the low-code Generative AI Studio, empowering developers with diverse options for their AI needs.
This announcement was made at the Google Cloud Executive Forum, a gathering of customers and partners held at the Bayview Campus in Mountain View. Previously, many of these advanced capabilities were only available to trusted testers, but now they are open to a wider audience.
The Codey model, introduced by Google as a code completion solution, is also making its debut in the Model Garden. Several Google partners have already started utilizing Codey, showcasing its value across various applications. For instance, GitLab is leveraging Codey’s capabilities in their “explain this vulnerability” feature. Canva, on the other hand, is utilizing Vertex AI’s translation feature to enhance support for their non-English speaking users, and they are now testing the PaLM model to transform short video clips into longer stories.
Typeface, a company focused on creative business use cases for generative AI, has also announced a partnership and deep integration with Google’s AI tools. By leveraging Google’s large language models, Typeface is launching as a plug-in in the Google Workspace Marketplace. Another partner, DataStax, has expanded the capabilities of its Astra DB database service on Google Cloud by adding support for vector search. This empowers AI models to use the service for long-term memory and retrieval.
Furthermore, Neo4j, a leading graph database, has integrated with Google Cloud’s generative AI features in Vertex AI. This integration enables users to transform unstructured data into knowledge graphs, providing the ability to query these graphs using natural language. Notably, Neo4j is also using Vertex AI to enrich these knowledge graphs in real-time. To ensure reliability and accuracy for enterprise customers, they have introduced the capability to validate responses from large language models against these enterprise knowledge graphs, preventing inaccuracies or false information.
Twilio, a renowned communications platform, has also expressed interest in exploring new Vertex-powered features. They are considering personalized “best next action” recommendations for contact center agents, which can greatly enhance customer service, as well as automating call summaries to streamline operations and improve efficiency.
Overall, these developments highlight the expanding capabilities of Google’s generative AI offerings and the strong partnerships formed to leverage these advanced technologies. By integrating cutting-edge AI models into their platforms and services, these partners can deliver innovative solutions that cater to diverse business needs, ultimately driving advancements in various industries.
What is Vertex AI?
During the Google I/O conference, Sundar Pichai, CEO of Alphabet, unveiled Vertex AI, an advanced managed machine learning platform designed for seamless deployment and maintenance of AI models. This innovative platform integrates the capabilities of AutoML and AI Platform, offering developers a unified API, client library, and user interface to streamline their machine learning workflows. With Vertex AI, businesses can harness the power of AI more efficiently and effectively.
Features of Vertex AI
1. Support for all open source frameworks
Vertex AI seamlessly integrates with popular open-source frameworks like TensorFlow, PyTorch, and scikit-learn. It also provides support for custom containers, allowing training and prediction using any ML framework.
2. Unified UI for the entire ML workflow
With Vertex AI, you have a unified user interface (UI) and API that brings together various Google Cloud services for building machine learning models. Whether you choose to use AutoML or custom code training, you can efficiently train and compare models. The central model repository makes it easy to deploy models to the same endpoints.
3. Pre-trained APIs
Vertex AI offers a range of pre-trained APIs for vision, natural language processing, video analysis, and more. These APIs can be easily integrated into existing applications, enabling you to add powerful AI capabilities without extensive development. You can also leverage these APIs to build new applications for translation, speech-to-text, and other use cases.
4. AutoML for tailored models
Developers can utilize AutoML in Vertex AI to train high-quality models that are specifically tailored to their business needs. The platform provides a central registry to manage and organize datasets, including those for vision-based and tabular data.
5. Integration with BigQuery and data-to-AI lifecycle
Vertex AI seamlessly integrates with BigQuery ML, allowing developers to create and execute machine learning models using standard SQL queries within familiar tools and spreadsheets. It also facilitates the smooth transfer of datasets from BigQuery to Vertex AI, enabling seamless integration across the entire data-to-AI lifecycle.
Vertex AI’s custom model tooling is designed to support advanced ML coding, offering a streamlined and efficient experience. It enables data scientists to train models with custom libraries using 80% fewer lines of code compared to other platforms, saving valuable time and effort.
The platform’s MLOps tools simplify the process of model maintenance, eliminating the complexities often associated with self-service model management. It provides seamless integration with ML pipelines and the Vertex Feature Store, allowing for easy serving, sharing, and utilization of ML features.
What sets Vertex AI apart is its accessibility to data scientists without formal ML training. It offers a comprehensive suite of tools that empower data scientists to manage their data, prototype models, experiment with different approaches, deploy models into production environments, interpret model results, and monitor performance—all within a single platform. This cohesive and user-friendly environment enables data scientists to maximize their productivity and make informed decisions throughout the entire machine learning lifecycle.