← All repos

generative-ai

Sample code and notebooks for Generative AI on Google Cloud, with Gemini Enterprise Agent Platform

GitHubdocs.cloud.google.com/gemini-enterprise-agent-platform/
agentsgcpgeminigemini-apigen-aigenerative-aigooglegoogle-cloudgoogle-geminilangchainlarge-language-modelsllmvertex-aivertexaivertex-ai-gemini-api
Browse cluster: Claude AI Agent Frameworks & MCP Tools →
2,179commits
312contributors
22languages

Tech stack & purpose

This repository contains sample code and notebooks for building generative AI applications on Google Cloud, specifically demonstrating the Gemini Enterprise Agent Platform. The project includes two detailed examples. The first, a Contract Compliance Multi-Agent Team built by Shubham Saboo and Eric Dong, demonstrates how to architect cross-language, auditable compliance agents using Python FastAPI with Google's Agent Development Kit (ADK) and a deterministic Go agent. It emphasizes using deterministic, policy-enforcement code for compliance checks rather than relying solely on language models, splitting responsibilities between a Python orchestration service that handles intake and extraction, and a Go A2A compliance agent that validates extracted fields through JSON-RPC. The system includes a polished web UI built with HTML5, CSS3, and vanilla JavaScript that lets contract reviewers run compliance audits and monitor the handoff between services. The second example, a Long Running AI Agent Team for New Hire Onboarding, demonstrates how to architect agents that handle time-agnostic, multi-day workflows. It uses Gemini Flash-Lite powered by the Google Agent Development Kit to guide new hires through a structured onboarding state machine. The project emphasizes durable, lightweight agent memory using enum-based state machines stored in SQLite rather than raw chat logs, paired with event-driven dormancy gates that pause execution during periods of external waiting (such as signature collection or hardware delivery). It includes a React-based live demo UI served by FastAPI that shows honest pause-and-resume behavior.

Community & reference links

Languages

Jupyter Notebook
83.1%
Python
7.3%
HTML
2.8%
JavaScript
2.7%
TypeScript
1.4%
SCSS
1.3%
Go
0.6%
CSS
0.3%
Shell
0.2%
Dart
0.1%
PLpgSQL
0.1%
Dockerfile
0.1%
C++
0.1%
CMake
0.0%
HCL
0.0%
Ruby
0.0%
Swift
0.0%
Makefile
0.0%
C
0.0%
Kotlin
0.0%
Objective-C
0.0%
Procfile
0.0%

Contributors (top 30 of 312)