Build Agentic AI & Multi-Agent System for Software Testing

Delve into the course structure and fundamentals of generative AI, AI agents, and multi-agent systems for software testing, including basics of lang chain doc parsing and vector stores.

Explore generative AI fundamentals, including large language models, GANs and VAEs, and how they generate text, images, or code, supported by weights, tokenization, and real-time web searching via AI agents.

Explore how ai agents empower large language models to think, decide, and act, using external tools to generate test cases.

explore the model context protocol from cloud anthropics and its mcp server, showing how ai agents interact with real-time data to automate browser tasks via playwright.

explore multi-agent systems, a team of ai agents that communicate to perform qa tasks such as scenario understanding, test case writing, and bug reporting.

Demonstrate practical multi-agent systems, with a researcher, coordinator/planner, and summarizer, as autogen and long chain orchestrate local models and data sources to deliver refined, multi-step insights.

Discover leading AI agent frameworks for building AI agents and multi-agent systems. Learn how LangChain and Autogen enable testing, RAG operations, and scalable AI applications.

explore running local large language models with olama to save money by avoiding api calls, install on macOS, Linux, or Windows, and choose models with embedding, vision, or tool support for internal testing.

Compare local ai models from 7b to 671b parameters, and assess memory, quantization, headcount, and hardware needs like gpu power and ram.

learn how to run local large language models with Ollama, list and download models (including 1.8 billion and deep seq variants), and generate code like selenium c-sharp dotnet offline.

Explore using Ollama models through a familiar GUI to run local large language models, switch between variants, perform inference, upload documents, and integrate vision and code generation.

Master the Ollama command line to remove, list, and show model details, compare its docker-like workflow, and prepare for using it locally with long-chain approaches.

Learn to start ollama as an api server, expose port 11434, and query a llama model via the generate endpoint using api requests to send prompts and receive responses.

Set up a Python environment, Visual Studio Code, and a virtual environment. Install Lang chain, Autogen, and Jupyter notebook to prepare for the Lang chain QA workflow using agents.

Install and configure a Jupyter notebook in Visual Studio Code, set the Python virtual environment kernel, write and run code cells, and document steps with markdown for a runnable runbook.

Explore Lang Chain 0.3 to 1.0 breaking changes, including package name updates, removed methods, and moved modules, as the course updates its source code for AI agents and multi-agent testing.

Explore LangChain basics and its ecosystem for building ai agents and qa testing, leveraging retrieval augmented generation, vector stores, and ai powered search with minimal code.

Learn to wire LangChain with a local Ollama LLM to build QA tools using multi-agent systems, guiding code setup, installation, and interactive testing in a notebook.

Interact with a local llama model via Ollama and LangChain by configuring the base URL locally at localhost:11434 and an 8 billion parameter model, then invoke and print the response.

Explore message streaming with LangChain by streaming the large language model's output in real time, using the stream method instead of invoke, and looping through tokens to display data incrementally.

Highlight LangChain v1.0 breaking changes, focusing on the single update: replace langchain dot prompts with langchain underscore dot prompts across code in this section.

Explore how prompt templates enable reusable, parameterized prompts for software testing with LangChain. See how this approach improves interactions with large language models for test generation and analysis.

Explore LangChain 1.x breaking changes, switching from lang chain prompts to lang chain underscore core prompts, and update your code to the new namespace for compatibility.

Master LangChain's chaining mechanism by turning a prompt template into a single chain that feeds data to the LLM, reducing invocations and enabling production-ready workflows with rags and documents.

Explore how chaining with output parsers in lang chain turns language model outputs into strings or JSON, enabling more reliable production-grade software testing workflows.

Learn to use Python to read local documents, split them into chunks, and store them for retrieval augmented generation to empower a large language model with a customized knowledge base.

Load document formats, chunk content, and prepare it for embedding and storage. Use LangChain and the unstructured PDF loader to read from a docs folder and build full file paths.

Load pdf documents into memory with an unstructured pdf loader using a file path, extend to combine documents, resolve packages like unstructured and pdfminer, and prepare chunks for language model.

Split large documents into 300-character chunks with 50-character overlap to balance efficiency, retrieval accuracy, and coherence, then store chunks in a vector store for Rag context to the language model.

Learn the LangChain 1.0 breaking change by switching from the old text splitter to the LangChain_text_splitter, and use text_splitters in plural to restore chunking.

Store document chunks in a vector data store to enable a Rag (retrieval augmented generation) system, using chroma vector stores, embedding functions, and LangChain.

Learn how to create document embeddings with a local model, using Namik embed text for embedding and Olama embedding when needed, and store the vectors in a Chroma vector store.

Adapt to LangChain 1.0 changes by using lang_chain_text_splitters, updating to lang_chain_olama.embedding for embeddings, and preparing for retrieval QA adjustments.

Learn how to store document chunks in a vector store using chroma via LangChain, with embeddings and a persistent qa db, preparing for retrieval with a retriever and retrieval qa.

Learn to retrieve documents from vector stores with retrieval QA by configuring a local language model, building a retriever from the DB, and composing a QA chain to fetch results.

Demonstrate data retrieval with a qa chain using a local large language model and a retriever to fetch data from the vector stores, featuring embedding and rag.

Learn how LangChain 1.0 deprecates retrieval QA and moves it to the LangChain Classic package, requiring code changes to use langchain_classic.chain for compatibility.

Learn to build ai agents and tools in LangChain for think, decide, and act with external tools, binding tools to llm for tasks like summarizing logs.

Build an artificial intelligence agent with tooling support using the Wikipedia tool in the LangChain ecosystem, initializing the agent and loading tools to fetch online data with local LLMs.

Understand breaking changes in LangChain 1.0 for AI agents, migrate code with create agent, and use Lang chain underscore classic for simpler integration.

Build a simple custom AI agent and tooling from scratch with LangChain, creating arithmetic tools and wiring them to an agent that executes add, subtract, and multiply.

Build genai and multi-agent system tools for software testing by orchestrating multiple tools in parallel, performing arithmetic, verification, and Jira ticket workflows with tool calling.

Create LangChain 1.0 agents using create_agents, replacing initialize_agent, and pass a model as LM. Use the invoke method with a human message array to call multiple tools.