Google’s generative AI support in Vertex AI is now generally available

Image credit: Michael Short/Bloomberg/Getty Images

Google today announced the general availability of generative AI support in Vertex AI, its machine learning platform. Based on Google’s models such as PaLM 2, Imagen, and Codey, Vertex AI provides text generation and classification, building multi-turn chat experiences like ChatGPT, and text embedding APIs for tasks such as semantic search and recommendation engines. Provides developers with access to PaLM’s capabilities for These models will be available as part of Google’s Vertex AI ‘Model Garden’, a collection of enterprise-ready foundational and task-specific models, and in the low-code Generative AI Studio.

The company made the announcement today in conjunction with a small gathering for customers and partners (the Google Cloud Executive Forum) at its Bayview campus in Mountain View. Until now, many of these features were only available to trusted testers.

It’s also worth noting that Google’s recently announced code completion model, Codey, and Chat model, PaLM, will debut in Model Garden.

Many Google partners are already leveraging Codey, for example GitLab uses Codey for its “Describe this Vulnerability” feature. Meanwhile, Canva is testing his PaLM model, which uses Vertex AI’s translation capabilities to better support non-English-speaking users and transforms short video clips into longer stories.

Typeface, which focuses on creative business use cases for generative AI, also today announced a partnership and deep integration with Google’s AI tools. The company leverages Google’s large language model and releases his Typeface as a plugin in the Google Workspace Marketplace. Meanwhile, DataStax announced that its Astra DB database service on Google Cloud now supports vector searches. This allows AI models to use the service for long-term memory.

Graph database Neo4j is also announcing new integrations with Google Cloud’s generative AI capabilities in Vertex AI. New here is the ability to transform unstructured data into a knowledge graph that users can query using natural language. The company also plans to use Vertex AI to power these knowledge graphs in real-time, which is important for enterprise customers, but to prevent hallucinations, the company also plans to combine responses from large-scale language models into these enterprise knowledge graphs. Add functionality to validate against graphs.

Twilio also announced today that it is exploring new capabilities powered by Vertex, including new personalized “best next action” recommendations for contact center agents and automated call summaries.