LangChain Runnables – Conveys Intelligent Automation

LangChainRunnables is a Python-based project showcasing five distinct LangChain workflows—Branch, Lambda, Parallel, Passthrough, and Sequence—using OpenRouter’s free API. It demonstrates AI-driven text processing tasks like generating facts, summarizing reports, creating notes and quizzes, responding to sentiment feedback, and crafting jokes.

---

🚀 What is a Runnable?

A Runnable is a core concept in LangChain used to define composable, reusable units of logic (chains, tools, models, etc.) that can be executed (or "run"). It abstracts components like LLMs, prompts, and retrievers into a unified interface.

Types of Runnables

1. Task-Specific Runnables

These are components designed to perform individual NLP tasks:

• Examples:
  • ChatOpenAI: Handles conversational LLM output.
  • PromptTemplate: Templates prompts for consistent input formatting.
  • Retrievers: Fetch relevant documents for context.

2. Runnable Primitives

These are combinators that build complex workflows from simpler tasks:

• RunnableSequence: Runs components in order, piping outputs to inputs.
• RunnableParallel: Executes components in parallel and merges results.
• RunnableBranch: Dynamically routes input based on conditions.
• RunnablePassthrough: Simply passes the input through to the next step.
• RunnableLambda: Allows custom Python logic inline.

---

🧠 LangChain Workflows

Each script in this repository showcases a different workflow pattern:

Workflow	Script	Description
Branch	runnable_branch.py	Classifies sentiment and tailors a response (e.g., “This is a beautiful phone”)
Lambda	runnable_lambda.py	Generates 5 interesting facts on a given topic (e.g., N8N)
Parallel	runnable_parallel.py	Creates notes and quizzes in parallel and merges the result
Passthrough	runnable_passthrough.py	Generates a joke based on input topic (e.g., programming)
Sequence	runnable_sequence.py	Produces a detailed report and a 5-point summary (e.g., AI ML jobs)

---

example workflow of runnable_passthrough project.

Passthrough: Generates a joke based on input topic {programming}. (Create a joke and count words in the joke using the NLP concept.)

🛠️ Technologies Used

• Python 3.11
• LangChain – Framework for building NLP workflows
• langchain-core – Core components for LangChain runnables
• openai – For interacting with OpenRouter APIs
• python-dotenv – Environment variable management
• pydantic – For structured output (used in branch sentiment classification)
• grandalf – For ASCII workflow visualization

---

✨ Features

• ✅ Fact generation, summarization, quiz creation, sentiment analysis, and joke generation
• 🌐 Free OpenRouter LLM integration
• 🧩 ASCII graph visualization of chains
• 🧾 Structured output parsing via Pydantic
• 🔐 Secure configuration with .env file for API keys
• Output and workflow visualization attached in file as a comment after code for quick and better understanding.

---

🔧 Installation

Clone the repository:

git clone https://github.com/vishal815/vishal815-LangChains-runnable-components-Conveys-intelligent-automation.git
cd vishal815-LangChains-runnable-components-Conveys-intelligent-automation

Steps for run project

Create a virtual environment:

python -m venv venv
venv\Scripts\activate

Install dependencies:

pip install -r requirements.txt

Create a .env file in the root folder and add your API key:

OPENROUTER_API_KEY=your-api-key-here

---

▶️ How to Use

Make sure your .env file is correctly set up.

Run any of the following Python scripts to explore the workflows:

python runnable_branch.py       # Sentiment-based feedback response
python runnable_lambda.py       # Topic-based fact generation
python runnable_parallel.py     # Note + quiz creation from text
python runnable_passthrough.py  # Joke generator
python runnable_sequence.py     # Report + summary creation

Each script will also render an ASCII graph to visualize how the workflow is structured.

---

👨‍💻 Author

• Name: Vishal Lazrus
• GitHub: vishal815

---

🤝 Contributing

Contributions are welcome!