Bringing the Power of AI to Your Application
summary
Covers a practical path to integrating OpenAI models (GPT‑4 family) with concrete techniques: 128k token context window and streaming vs batch tradeoffs, prompting patterns like examples, instruction prompts, Chain of Thought, JSON output and retrieval augmented generation with embeddings stored in vector DBs. Describes real prototypes—Copilot for Docs plus an offline evaluation that regenerates code across hundreds of thousands of Python functions and reruns unit tests—and shows results are measured by A/B telemetry where latency sometimes beats raw accuracy.
AI-written summary · about
chapters (10)
- 0:00Introduction to AI talk
- 1:10About GitHub Next team
- 2:36AI integration made simple
- 4:20Large language models overview
- 11:28Prompt engineering techniques
- 23:11Retrieval augmented generation
- 28:49Measuring AI project success
- 35:17User experience with AI
- 41:59Should you use AI?
- 43:26Q&A and closing remarks
related talks
32mHow to Make AI Work for You: A Deep Dive Guide into ChatGPT for IT Professionals
59mKeynote: The past, present, and future of AI for application developers
40mDavid Fogl: Leveraging Generative AI for Enhanced Software Testing
27mBuilding AI applications: Successes and Lessons
46mUsing AI in Software Design: How ChatGPT Can Help With Creating a Solution Arch.
24mInfuse Generative AI in your apps using
58mKeynote: The past, present, and future of AI for application developers
01h 01mUnlocking the Potential of AI