Feature availability
Build an agent system with a complete AI stack
Vertex AI eliminates the trade-off between accessing frontier models and leveraging a world-class AI development platform. Building advanced agents requires more than just a model, so Vertex AI provides a complete agent stack designed to manage operational, governance, and global complexity.
Build intelligent agents quickly. take advantage of Vertex AI Agent Builder Stack— including it agent development kit– Move multi-step agent workflows from prototype to large-scale production.
Extend your agents with fully managed services. use agent engine Deploy and scale agents in serverless environments. Using memory banks and sessions, Claude-powered agents maintain persistent long-term context across multiple interactions, resulting in a more natural user experience.
Improve performance and reduce costs. Maintain consistent performance during peak demand by using Provisioned Throughput to reserve dedicated capacity at a fixed cost. To further improve efficiency, take advantage of features such as: Claude’s global endpoint, prompt cache Flexible time-to-live (TTL) allows batch prediction, Web search (preview),and 1M Token Context Window (Preview) Claude works for 4.6.
Run on optimized infrastructure. Vertex AI’s infrastructure is purpose-built to handle the computing demands of frontier models, reduce latency, and increase efficiency.
Governance and security at scale: With integrated governance controls, agents always perform their tasks and their actions are recorded. Protect your complete AI stack, including models and agents, with basic security and data residency controls from Google Cloud, protecting against emerging AI threats like prompt injection and tool poisoning. model armorand become an expert AI protection in security command center AI Prioritize risks and detect threats.
