Simple AI-Agent Web App
Budget: ₹1,000 – ₹2,500 INR
I need a bare-bones prototype that proves an AI agent can accept any text task from a user, execute it autonomously, and display the result—no conversational fluff, just action and output.
Tech stack
• Python with FastAPI (please stick to this choice)
• OpenAI API or a comparable LLM endpoint to power the agent logic
• Lightweight HTML template or bare React/Vue front end is fine; keep the UI minimal so users can type the task and immediately see the result.
Core requirements
1. One input box where the user writes a task.
2. The backend passes that task to the agent; the agent decides the steps and completes the task.
3. Return the final result straight to the browser.
4. Everything should run locally after a simple pip install ‑r requirements.txt and uvicorn main:app. No database, no Docker on this first cut.
Acceptance criteria
• FastAPI service starts without errors, exposes one route for the UI and one for processing.
• Agent code is clearly separated so future skills (data processing, web scraping, file manipulation, etc.) can be added easily.
• README with setup and usage instructions.
• Delivered as a single Git repo or zipped folder.
This first version is an MVP, so I’ve kept the budget lean (₹2,000 fixed). Impress me with clean, well-commented code and a smooth local run, and there will be follow-on work to expand the agent’s capabilities.
Tech stack
• Python with FastAPI (please stick to this choice)
• OpenAI API or a comparable LLM endpoint to power the agent logic
• Lightweight HTML template or bare React/Vue front end is fine; keep the UI minimal so users can type the task and immediately see the result.
Core requirements
1. One input box where the user writes a task.
2. The backend passes that task to the agent; the agent decides the steps and completes the task.
3. Return the final result straight to the browser.
4. Everything should run locally after a simple pip install ‑r requirements.txt and uvicorn main:app. No database, no Docker on this first cut.
Acceptance criteria
• FastAPI service starts without errors, exposes one route for the UI and one for processing.
• Agent code is clearly separated so future skills (data processing, web scraping, file manipulation, etc.) can be added easily.
• README with setup and usage instructions.
• Delivered as a single Git repo or zipped folder.
This first version is an MVP, so I’ve kept the budget lean (₹2,000 fixed). Impress me with clean, well-commented code and a smooth local run, and there will be follow-on work to expand the agent’s capabilities.