What is the ADK? AI Agents for real time market analysis, and Top 10 opensource Github repos

In this blog, we explore ADK, AI Agents for real time market analysis and top 10 open source github repositories for building RAG and AI Agents.

In this blog, I decided to get together a quick overview of what ADK is and the types of agents you can build with it!

ADK is Google’s flexible, modular framework for building and orchestrating AI agents, tightly integrated with the Google ecosystem and optimized for Gemini models.

With ADK, you can build everything from simple Gemini-powered assistants to complex, multi-agent workflows.

Here's a quick overview of the agents you can build:

LLM Agents

These are your reasoning engines, powered by LLMs for natural language understanding, decision-making, and tool use.

Example: Smart Email Assistant

class EmailReplyAgent(LlmAgent):

    def __init__(self):

        super().__init__(

            name="email_reply_agent",

            instructions="You're an assistant that drafts professional email replies.",

            tools=[GrammarCheckTool(), SummarizerTool()]

        )

Use Cases:

• AI customer support

• Email drafting

• Research summarization

• Tool-using agents (e.g., search + act)

Workflow Agents

Deterministic control agents that organize how other agents run:
SequentialAgent: Executes agents in specific order

Example: Onboarding Workflow

onboarding_agent = SequentialAgent(
    name="onboarding_sequence",
    sub_agents=[CollectUserInfoAgent(), SendWelcomeEmailAgent(), ScheduleIntroCallAgent()]
)

Use Cases:

• User onboarding

• Multi-step data pipelines

• Any fixed-sequence task

LoopAgent: Repeats a set of agents based on a condition

Example: Code Review Loop

loop_review_agent = LoopAgent(
    name="code_review_loop",
    sub_agents=[GenerateFixSuggestionAgent(), ApplyFixAgent()],
    max_iterations=3
)

Use Cases:

• Iterative refinement (e.g., code or content)

• Retry loops

• Model fine-tuning steps

ParallelAgent: Runs agents concurrently to save time
Example: Multi-Channel Content Generator

parallel_content_agent = ParallelAgent(
    name="content_creator_parallel",
    sub_agents=[BlogWriterAgent(), LinkedInPostAgent(), TweetAgent()]
)

Use Cases:

• Multi-format content generation

• Parallel data collection

• Independent tool execution

Custom Agents

For when you need total control. Define your own orchestration logic by extending the BaseAgent class.

Example: Conditional Routing Agent

class ConditionalAgent(BaseAgent):
    async def run(self, input_data):
        if input_data["priority"] == "high":
            return await HighPriorityAgent().run(input_data)
        else:
            return await LowPriorityAgent().run(input_data)

Use Cases:

• Conditional routing

• Multi-path workflows

• Domain-specific decision trees

  

  
  

ADK represents the next generation of AI tooling. If you’re looking to build with LLMs beyond chatbots, this is a framework worth exploring.

Github repo: https://github.com/google/adk-python

Sample ADK: https://github.com/google/adk-samples

---

AI Agents for real time market analysis and executing trading strategy

I recently built a multi-agent AI system that helps automate market analysis and trading signals, giving users smarter real-time insights while keeping them in control of AI recommendations.

Why use multi-agent AI for trading?
Multi-agent systems break down complex tasks into specialized agents.

Instead of a single AI trying to do everything, each agent focuses on a specific function, analyzing stock trends, tracking news, generating buy/sell signals, and even optimizing execution strategies.

Here’s how it works:
→ Analyzes real-time market data and trends 📊
→ Generates buy/sell signals for stocks 📈
→ Lets users approve or reject trades before execution
→ Once approved, executes trades using a smart strategy 🚀

This is a step towards the future of AI-assisted trading, where technology enhances decision-making, but you stay in control.

Watch the full video here.

Link to the github repository here.

---

Top 10 RAG and AI Agents Open Source Github repositories

If you're into AI agents, Retrieval-Augmented Generation (RAG), or workflow automation, you will find these repos helpful:

🔹 OpenManus: No need for an invite! This open-source library lets you use ManusAi features right away. GitHub link.

🔹 GenAI Agents by Nir Diamant: if you want to understand how AI agents work, this repo breaks down different techniques and approaches. GitHub link.

🔹 bRAG-Langchain: A deep dive into Retrieval-Augmented Generation (RAG) with hands-on examples. If you're building with RAG, this is a must-watch. GitHub link.

🔹 Awesome-LLM-as-a-Judge by Dawei Li and his folks: A well-curated collection of research on using LLMs to evaluate AI-generated content. Great for AI benchmarking. GitHub link.

🔹 Awesome-llm-apps by Shubham Saboo: A curated collection of awesome LLM apps built with RAG and AI agents. GitHub link.

🔹 FlowiseAI (YC S23), Drag & drop UI to build your customized LLM flow. Github link.

🔹 Dify is an open-source LLM app development platform: https://github.com/langgenius/dify

🔹 Helm is an open-source library for Holistic Evaluation of Language Models: https://github.com/stanford-crfm/helm

🔹 Langflow is a powerful tool for building and deploying AI-powered agents and workflows https://github.com/langflow-ai/langflow

🔹 LLM4Annotation is an open-source library for Large Language Models for Data Annotation and Synthesis: https://github.com/Zhen-Tan-dmml/LLM4Annotation

If you're building with AI, these repos can save you time and help you build faster.

If you know any other helpful open-source AI projects, drop them in the comments, I’d love to check them out! 🚀