Building Agents

A collection of small, self-contained agentic AI projects, each exploring a different agent pattern and SDK.

Modules

foundations/

A personal-assistant chatbot that answers questions as "you" on a website (e.g. for recruiters or clients).

• Uses the OpenAI Chat Completions API with function calling.
• Builds its persona from me/summary.txt and me/linkedin.pdf.
• Exposes two tools: record_user_details (captures a visitor's email/notes) and record_unknown_question (logs questions it couldn't answer), both delivered via Pushover push notifications.
• Served through a Gradio chat UI.

Run it with:

cd foundations
python app.py

Requires PUSHOVER_TOKEN and PUSHOVER_USER env vars, and an OPENAI_API_KEY.

deep_research/

A multi-agent research pipeline built on the OpenAI Agents SDK. Given a topic, it plans web searches, executes them, synthesizes a report, and emails it out.

• planner_agent.py — plans a set of web searches for the query (WebSearchPlan).
• search_agent.py — runs each search with WebSearchTool and returns a concise summary.
• writer_agent.py — synthesizes all search summaries into a long-form markdown report (ReportData).
• email_agent.py — converts the report to HTML and sends it via SendGrid.
• research_manager.py — orchestrates the agents end-to-end (plan → search → write → email) and streams progress updates.
• deep_research.py — Gradio UI entry point.

Run it with:

cd deep_research
python deep_research.py

Requires OPENAI_API_KEY and SENDGRID_API_KEY env vars.

Setup

This project uses uv for dependency management.

uv sync

Create a .env file in the project root with the API keys referenced above.