A curated list of code samples and tutorials for building agents on Gemini Enterprise Agent Platform (previously Vertex AI).
- Onboarding Guide - A quick-start guide to accelerate your onboarding to the platform.
The state-of-the-art intelligence powering the entire platform.
- Gemini API - Access Google's most capable generative ai models.
- Veo (Video) - High-definition video generation and editing.
- Lyria (Music) - World-class music and audio generation.
- Nano Banana (Image) - Fast, high-quality image generation.
- Model Garden - Discover and deploy 150+ partner and open-weight models (including Claude).
🛠️ Build - Turning code into agents
- Development Framework
- Agent Development Kit (ADK) - Open-source, model-agnostic framework for building production agents.
- Agents CLI - Command-line tool for scaffolding, testing, and deploying agents.
- Builders & Discovery
- Agent Studio - Low-code visual interface for rapid agent prototyping and deployment.
- Agent Garden - Discover, customize, and deploy pre-built agent templates.
- Tools & Data Integration
- Grounding - Connect models to real-world data and Google Search for factual accuracy.
- RAG Engine - Managed Retrieval-Augmented Generation for enterprise datasets.
- Agent Search (previously Vertex AI Search, Generative AI App Builder) - Semantic search engines tailored for your data.
- Vector Search 2.0 - High-scale, low-latency managed vector database.
Open standards to help agents communicate, perform actions, and generate interfaces.
- Agent2Agent (A2A) - Secure protocol for multi-agent coordination and handoffs.
- Model Context Protocol (MCP) - Universal standard for connecting models to external tools.
- Agent-to-UI (A2UI) - Protocol for dynamic, agent-driven user interface generation.
- Agent Payments Protocol (AP2) - Secure standard for automated financial transactions by agents.
- Universal Commerce Protocol (UCP) - Unified standard for managing e-commerce and retail operations.
📈 Scale - Reliability over time
- Agent Runtime (previously Agent Engine, Reasoning Engine) - Managed platform for hosting and serving agentic code.
- Agent Sessions - Built-in state management for multi-turn, long-running interactions.
- Agent Memory Bank - Persistent, long-term semantic memory for personalized agent experiences.
- Agent Sandbox (Code Execution) - Secure, isolated environment for code execution and computer use.
🔐 Govern - Trust as a feature
- Agent Gateway - Central control point for authentication, authorization, and rate-limiting. (Private Preview)
- Agent Identity - Unique credentials for agentic identities.
- Agent Policies - IAM Policies specifically for agents. (Private Preview)
- Agent Registry - Enterprise catalog for managing and auditing approved tools and agents.
- Model Armor - Real-time protection against prompt injections, PII leaks, and toxic content.
⚙️ Optimize - Evaluation and enhancement
- Agent Evaluation - Framework for testing complex reasoning and tool-use accuracy.
- Agent Simulation - Generate synthetic user interactions to stress-test agent behavior.
- Agent Observability - Forensic visibility into agent thought processes and tool calls.
- Agent Optimizer - Automate the improvement of agent prompts based on failure analysis.
Deep dives into specific platform capabilities and use cases.
- Intro to Gemini 3.1 Pro
- Intro to Gemini 3 Flash
- Intro to Gemini 3.1 Flash-Lite
- Intro to Batch Prediction
- Intro to Code Execution
- Intro to Computer Use
- Intro to Agentic Vision
- Intro to Context Caching
- Intro to Live API
- Intro to Supervised Fine Tuning
- Intro to Evaluation
- Document Processing
- Spatial Understanding
- YouTube Video Analysis
- Intro to Prompt Attacks and Mitigation
- Intro to LangGraph
Embeddings & Vector Search
- Getting Started with Text Embeddings + Vector Search
- Introduction to Multimodal Embeddings
- Introduction to Vector Search 2.0
- Hybrid Search Tutorial
This is not an officially supported Google product. This project is not eligible for the Google Open Source Software Vulnerability Rewards Program.