Enhanced AI Agent Builder: Google Cloud’s Latest Updates for Improved Performance and Efficiency

Google Cloud recently unveiled a significant update to its Vertex AI platform. The update aims to retain AI developers on the platform by providing a comprehensive solution for concepting, designing, building, testing, deploying, and modifying AI agents for enterprise applications.

The new features introduced today include enhanced governance tools for enterprises, expanded agent creation capabilities with minimal code, streamlined context management layers, simplified one-click deployment, managed services for scaling production, evaluation, and agent identification support.

Agent Builder, which was initially launched during the previous year’s Cloud Next event, offers a no-code platform for enterprises to create agents and seamlessly integrate them with orchestration frameworks like LangChain.

Google’s Agent Development Kit (ADK) allows developers to build agents in less than 100 lines of code and can now be accessed through Agent Builder, further streamlining the agent development process.

“These new capabilities highlight our dedication to Agent Builder, simplifying the agent development process to cater to developers regardless of their chosen tech stack,” said Mike Clark, Director of Product Management at Vertex AI Agent Builder.

Accelerate Agent Development

Google emphasizes that enterprises can now incorporate orchestration seamlessly during agent construction with the new features of Agent Builder.

“The process of building an agent from concept to a functional product involves intricate orchestration,” explained Clark.

The new capabilities introduced with the ADK include:

• State-of-the-art context management layers such as Static, Turn, User, and Cache layers for enhanced control over agent context

• Prebuilt plugins with customizable logic, including a plugin that enables agents to self-heal by retrying tasks with a different approach upon failed tool calls

• Additional language support in ADK, now including Go alongside Python and Java

• One-click deployment through the ADK command line interface for seamless transition from local environment to live testing

Enhanced Governance

Enterprises demand high accuracy, security, observability, and auditability for their production-grade AI agents.

Google now provides observability features through the Agent Engine managed runtime dashboard, enabling cloud-based production monitoring to track token consumption, error rates, and latency. Enterprises can visualize agent actions and address issues within this observability dashboard.

The new Evaluation Layer within Agent Engine assists in simulating agent performance across various user interactions and scenarios.

Additional governance features include Agent Identities, Model Armor for enhanced security, and Security Command Center for threat detection and inventory management.

Competition and Innovation in Agent Building

Various model providers offer platforms for agent creation, with the competition revolving around the rapid introduction of new tools and features.

Google’s Agent Builder competes with OpenAI’s Agent Development Kit and the recently announced AgentKit, as well as Microsoft’s Azure AI Foundry and AWS’s agent builders on Bedrock. Google aims to gain a competitive edge through its suite of new features.

Companies with their own agent libraries, along with enterprise service providers, are vying for developer attention to build AI agents within their platforms. Attracting and retaining developers within their ecosystems is a key focus for tech companies, emphasizing features that simplify agent creation and governance.