Keynote - The AI Playbook For Companies
summary
Four concrete plays decide whether an AI effort works: control — treat prompts as code, version test and guard them so outputs stop being generic; context — use RAG and MCP servers to prioritize local docs and preserve metaphors and dialects; specialization — fine‑tune with PEFT and LoRA to add skills without catastrophic forgetting; and agency — build agents that hit LinkedIn APIs, scrape press pages, draft emails and log to CRM so the system executes outcomes not just answers. Merchandise sales and a No Starch Press book fund charity.
AI-written summary · about
chapters (15)
- 0:00Welcome
- 1:26Sponsors
- 3:53Conference logistics
- 6:22This Dot Labs intro
- 8:48AI playbook discussion
- 9:29Human wisdom in AI
- 12:17Four AI plays
- 13:21Play 1: Prompt engineering
- 15:46Play 2: RAG and MCP servers
- 17:51AI localization
- 19:18Play 3: Fine-tuning
- 20:42Cost of perfection
- 21:27Play 4: AI agents
- 22:30Answers to outcomes
- 22:52Wisdom as advantage
related talks
38mKeynote: Thriving in the Age of AI
49mKeynote: Operations and AI with Jen Krieger
30mKeynote: Fireside Chat
32mKeynote: Will the Real Autonomous Agent Please Stand Up
33mKeynote: Boring Wins in AI Agents 🔥 Hard-Earned Lessons
01h 03mKeynote: Red, Blue, and Purple AI
01h 00mKeynote Panel AI from Investment perspective
39mKeynote Panel State of AI in the Enterprise