Google Cloud has introduced a major update in an effort to keep AI developers on its Vertex AI platform to imagine, design, build, test, deploy and adapt AI agents in business use cases.
The new features, announced today, include additional enterprise governance tools and expanding capabilities to create agents with just a few lines of code, work faster with state-of-the-art context management layers and one-click deployment, as well as managed services for scaling production and evaluation, and support for identifying agents.
Agent builder, released last year at its annual Cloud Next event, offers a no-code platform for enterprises to create agents and connect them to orchestration frameworks such as LangChain.
Google’s Agent development kit (ADK), which allows developers to build agents “in less than 100 lines of code,” can also be accessed through Agent Builder.
“These new capabilities underscore our commitment to Agent Builder and simplify the agent development process to meet developers where they are, no matter which tech stack they choose,” said Mike Clark, Director of Product Management, Vertex AI Agent Builder.
Build agents faster
Part of Google’s pitch for Agent Builder’s new features is that companies can orchestrate as they build their agents.
“Building an agent from a concept to a working product requires complex orchestration,” said Clark.
The new capabilities included with the ADK include:
SOTA context management layers, including static, turn, user, and cache layers, so enterprises have more control over the agents’ context
Pre-built plugins with customizable logic. One of the new plugins allows agents to recognize failed tool calls and fix themselves by rerunning the task with a different approach
Additional language support in ADK, including Go, in addition to Python and Java, launched with ADK
One-click deployment via the ADK command-line interface to move agents from a local environment to live testing with a single command
Administrative layer
Businesses require high accuracy; security; observability and controllability (what a program did and why); and drivability (control) in their production-grade AI agents.
While Google had observability features in the local development environment at launch, developers can now access these tools through the Agent Engine-managed runtime dashboard.
The company said this provides cloud-based production monitoring to track token consumption, error rates and latency. Within this observation dashboard, companies can visualize the actions agents are taking and reproduce any issues.
Agent Engine will also have a new evaluation layer to “simulate agent performance across a wide range of user interactions and situations.”
This level of government also includes:
Agent identities which Google said “give agents their own unique, native identity within Google Cloud
Model Armor, which would block quick injections, tool screen calls, and agent responses
Security Command Center, so administrators can build an inventory of their agents to detect threats such as unauthorized access
“These native identities provide a deep, built-in layer of control and a clear audit trail for all agent actions. These certificate-backed identities further strengthen your security because they are impersonable and directly tied to the agent lifecycle, eliminating the risk of dormant accounts,” said Clark.
The battle between agent builders
It’s no surprise that model providers are creating platforms to build agents and deploy them into production. The competition lies in the speed at which new tools and features are added.
Google’s Agent Builder competes with OpenAI‘s open source Agent development kitwhich allows developers to create AI agents using non-OpenAI models.
In addition, there is the recent AgentKit announcedwhich features an Agent Builder that allows companies to easily integrate agents into their applications.
Microsoft has its Azure AI Foundrylaunched around this time last year for creating AI agents, and AWS also offers agent builders on the Bedrock platform, but Google hopes a range of new features will give it a competitive advantage.
However, it’s not just companies with their own models that are pushing developers to build their AI agents within their platforms. Any business service provider with an agent library also wants customers to create agents on their systems.
Capturing developers’ interest and keeping them in the ecosystem is now the big battle among tech companies, with features to make building and operating agents easier.
#Google #Cloud #updates #Agent #Builder #observability #dashboard #faster #build #deployment #tools