How to work with generative AI in JavaScript
summary
Shows working JavaScript examples: call Gemini 2.0 Flash with nine lines of SDK code, create text-embedding-004 vectors (768 dimensions) and use cosine similarity for RAG, store and sort vectors in Astra DB vector index, and stream tokenized responses via generateContentStream to improve perceived latency. Recommends prompt params (temp 0.2, topP 0.95, topK 30), demonstrates a conference-talk search demo that answers queries, and builds simple agents with JSON-schema tool defs — an agent loop ~20 lines that calls functions and returns results.
AI-written summary · about
chapters (35)
- 0:00Introduction
- 1:02Why generative AI
- 2:45Working with generative AI
- 4:47Artificial Analysis
- 6:17Google Genai
- 7:25Demo application
- 9:12Talk to it
- 9:40Tokens
- 12:09Autocomplete
- 12:56TopK
- 17:19Prompt Engineering
- 19:22ZeroShot Prompting
- 20:52Example Prompt
- 21:37System Prompt
- 22:22Hard Challenges
- 25:26Chat
- 27:35Context
- 29:05Retrieval
- 30:51Vector search
- 35:11Vector sort
- 35:57Demo
- 38:47User experience
- 39:58Streaming responses
- 41:48Streaming results
- 43:23Defining agents
- 44:28Defining tools
- 45:23Defining functions
- 46:59Using functions
- 48:07The tool loop
- 50:13The bot
- 54:35The libraries
- 55:43Visual builders
- 56:38Browser models
- 57:24Gemini Nano
- 58:26Conclusion
related talks
01h 48mBuilding Conversational Agents
02h 44mUsing Generative AI with Java: Beyond Simple Chatbots
49mLet's build an AI agent
24mEmpowering Vue Apps with AI
01h 45mAI Engineering with the Google Gemini 2.5 Model Family
21mUnlocking Client-Side AI in Your Web App
22mClient side Web AI Agents: Building smarter user experiences for future agentic ...
44mHands-on-lab: It's all about going hands-on with Generative AI!