Building Generative AI Applications in Go- Gari Singh, Google (Part 1)
summary
Claim: Go now has a practical ecosystem to build and deploy generative AI apps end to end. The demo uses the Google Go SDK to call hosted Gemini models, swaps models via LangChain Go, serves local open models with AMA (running lightweight Gemma/llama variants on a Chromebook CPU), and embeds flows and prompt templates with Firebase gen kit. A Kubernetes-hosted chat app is shown working, streaming tokenized responses, and local model serving and tool chaining all run successfully in the demos.
AI-written summary · about
chapters (13)
- 0:00Introduction to AI in Go
- 3:43Understanding LLMs
- 7:25Generative AI use cases
- 10:14Building with Gemini API
- 15:35Prompt engineering basics
- 20:28Prompting for context
- 23:21LLM application stack
- 26:05Orchestration with LangChain
- 32:42Serving models locally
- 38:49Running Gemma locally
- 46:17Introduction to Genkit
- 50:25Genkit components
- 54:39Genkit in practice
related talks
28mBuilding AI Applications in Go
11mapproach to generative AI
43mAI and Go: Opportunities and Challenges Panel Discussion
31mAI-Driven Development in Go
20mA fireside chat
49mAbracadabra! Bring Your Ideas to Life with Google AI Studio
29mGenerative AI: A Google Perspective
29mThe Go-to Language for AI: Opportunities and Challenges