The march toward agent enterprises continues as companies fight to keep developers on their platforms throughout the agent lifecycle.
Google Cloud Vertex has updated its Agent Builder on AI, introducing additional governance tools for enterprises and enhanced capabilities to create agents with just a few lines of code.
agent builder, Released last year During its annual CloudNext event, Code provides a platform for enterprises to create agents and connect them to orchestration frameworks such as Lingchain. Google’s Agent Development Kit .
The new updates include advanced context management layers and features for one-click deployment as well as managed services for production and evaluation, and managed services to support agent identification, as well as features for rapidly building agents with one-click deployment.
“These new capabilities underscore our commitment to Agent Builder, and simplify the agent development process to meet developers, as far as their tech stack choices,” said Mike Clark, director of product management, Vertex AI Agent Builder.
Building agents fast
Part of Google’s pitch for the new Agent Builder features is that enterprises can cook in orchestration as they build their own agents.
“Building an agent from a concept to a working product involves complex orchestration,” Clark said.
New capabilities, which ship with the ADK, include:
SOTA context management layers including static, queue, user and cache layers so enterprises have greater control over agents’ contexts
Prebuilt plugins with custom logic. A new plugin allows agents to recognize failed tool calls and “self-heal” by retrying tasks with different approaches.
Additional language support in the ADK, including GO, along with Python and Java, which launched with the ADK
One-click deployment via the ADK command-line interface to transfer agents from the local environment to testing directly for testing with a single command
Governance layer
Enterprises want agents working for them, performing well, performing well and being able to control if something goes wrong.
While Google had Observable features in the native development environment at launch, developers can now access these tools through the Agent Engine’s managed runtime dashboard. The company said it brings cloud-based production monitoring to track token consumption, error rates and delays. Within this observational dashboard, enterprises can visualize agents’ actions and reproduce any issues.
The agent engine will also have a “new evaluation layer to help simulate user interactions and agent performance in a wide array of situations.”
This governance layer will also include:
Agent Identity What Google said “gives agents their own unique, local identity in the Google Cloud
Model Armor, which will prevent immediate injection, screen tool calls and agent responses
Security Command Center, so admins can inventory their agents to detect threats such as unauthorized access
“These native identities provide a deep, built-in layer and a clear audit trail for all agent actions. These certificate-backed identities further strengthen your security because they cannot be impersonated and are tied directly to the life of the agent, eliminating the risk of inactive accounts,” said Clark.
Battle of Agent Builders
It’s no surprise that model providers develop platforms for building agents and bringing them into production. The competition lies in how quickly new tools and features are added.
Google’s Agent Builder competes Open Eyeof open source Agent Development Kitwhich enables developers to create AI agents using nonopinion models. In addition, recently Announced Agent Kitwhich features an agent builder that enables companies to easily integrate agents into their applications. AWS Also offer Agent builders on its bedrock The platform
However, it’s not just companies with their own models that court developers to build their own AI agents into their platforms. are Any enterprise service provider with an agent library also wants clients to create agents on their systems.
Capturing developer interest and keeping them in the ecosystem is now a major battle between tech companies, with features to facilitate building and governing agents.