Connecting an LLM to your own data by ingesting documents, and retrieving relevant context to generate accurate, grounded answers with Google ADK, Vertex AI and Twilio. This approach avoids hallucinations and enables real-world AI applications for businesses.
You’ll see a working example of a RAG agent that can answer questions based on private documents.
And as an extension, how the same system can be connected to a real phone number with Twilio so you can call the agent on the phone and ask questions in real time.
This talk was presented at WarsawJS and ViennaJS.
Topics covered:
- Retrieval-Augmented Generation (RAG) explained
- Why prompt-only approaches fail (context limits)
- Building AI agents in JavaScript with Google ADK
- Document ingestion and semantic retrieval with Vertex AI
- FunctionTool pattern and clean architecture
- Voice AI integration (phone-call agent)
- Real-world use cases for developers and businesses
Full code example: https://github.com/dnlrbz/adk-rag-agent
Download
0 formats
No download links available.
JavaScript AI Agents: RAG + Google ADK + Phone Call Demo | NatokHD