This repository contains a collection of Agentic AI workflows built using structured multi-agent architectures. These projects explore how LLM planning, tool usage, reflection loops, evaluation feedback, and continuous improvement can be designed in real-world automated systems.
Agentic AI goes beyond simple prompt-based responses. These projects follow a systematic agent design workflow:
| Stage | Purpose |
|---|---|
| Initial Planning & Task Decomposition | Break down a user problem into logical steps |
| Reflective Design | Think before acting – evaluate steps and plan improvements |
| Tool Use (Functions & External APIs) | Use tools such as databases, file systems, web search, and compute functions |
| Evals & Error Analysis | Use feedback from LLM self-evaluation and/or external evaluators |
| Iterative Refinement | Improve responses using refinement loops |
| Impact Measurement | Show how structured planning improves output quality and reliability |
| S.No | Project Name | Description | Key Concepts |
|---|---|---|---|
| 1 | Interactive Chart Generator | Generates visual insights from CSV files and refines based on feedback | File Parsing, Visualization, Reflection |
| 2 | SQL Query Generation Agent | Validates SQL queries from DB schemas with reflection loops | DB Tools, Planning |
| 3 | Tool Toggle Agent | Chooses and toggles tools based on task goals | Decision Routing |
| 4 | Content Writing Agentic AI | Generates and refines content using constructive feedback | Reflection, Evals |
| 5 | Research Agentic AI | Researches topics using external tools and citation refinement | Web Tools, Self-Correction |
| 6 | Startup Builder | Analyses and idea communicates with 6 different agents and gives the market analysis, gap analysis, gap, scope, mvp features | Multi-Agent Workflow |
| 7 | Portfolio Generator | Analyzes the resume, parses it, generates a portfolio website based on it | Resume Parsing, Website Generation |
| 8 | Stock Data Visualizer | Retreives Stock information using Yahoo finance and plots a graph based on the data. Refines the charts. Uses 5 different agents | Stock Data, Visualization, Reflection |
Before running any project in this repository, create a .env file in the project root directory and add the following API keys:
OPENAI_API_KEY=your_openai_api_key
ANTHROPIC_API_KEY=your_anthropic_api_key
TAVILY_API_KEY=your_tavily_api_key
DLAI_TAVILY_BASE_URL=https://api.tavily.com- Python
- AISuite
- LLM APIs
- Plotly / Matplotlib
- SQLite/Postgres
- Docker (optional)
agenticAIProjects/
- charts_agent
- sql_agent
- tool_toggle_agent
- content_writer_agent
- research_agent
Pull requests are welcome.