Uploaded March 2025 | Updated September 2026, 2 weeks ago

---

While you can include a 1,000-page PDF file in a prompt request, the token usage is going to be extremely expensive, especially if you are using highly capable generative models like O1, Gemini Pro, or Sonnet, which might cost you a few dollars per request. By using RAG, you can reduce the cost by almost 100 times, even more so if you implement the correct RAG chunking strategy.

📋 Source Code (So you don't have to write all the code yourself)
➡️ patreon.com/posts/123189712

🖥️ More on PydanticAI: https://ai.pydantic.dev/

📃 LanceDB Documentation: lancedb.github.io/lancedb

💻 Also checkout:
➡️ Getting Started With PydanticAI (youtu.be/H63RbTuHTy0)
➡️ Create An Intelligent Web Scraping Agent With PydanticAI | Complete Tutorial (youtu.be/7tXSk31d9X0)
➡️ Google Sheets AI Agent With PydanticAI | Complete Tutorial (youtu.be/tURxphunUyk)

💖 Show Your Support
☕ Paypal: paypal.me/jiejenn/5
☕ Venmo: @Jie-Jenn
🌳 Patreon: patreon.com/JieJenn
✉️ Business Inquiring: YouTube@LearnDataAnalysis.org

00:00 - Intro to RAG
02:57 - Chunking strategy types
09:37 - Install Python dependencies
10:00 - Set up RAG databse with LanceDB
17:20 - Test RAG database with PydanticAI Agent

#pydanticai #aiagent #agenticai