How to Really Scale AI: Combining Traditional and Agentic Automation
summary
Claims that scaling fails for process reasons not model limits and shows a concrete hybrid build: SEO research automation that pulls Google Sheets Google Analytics and web results, a RAG‑style interlinking extractor over the CMS, and a copy agent version of a custom GPT wired into those flows so a single form triggers end‑to‑end content production. Implementation uses Make platform, OpenAI and RAG techniques, reports ~300 AI workflows and 100+ agents in production, and demonstrates a working system that moves projects past pilots.
AI-written summary · about
chapters (14)
- 0:00Introduction
- 1:10Scaling AI challenges
- 2:37Right AI use cases
- 5:02Seven AI problem types
- 12:55Combining AI use cases
- 13:58Content creation scaling
- 22:47Traditional automation foundation
- 32:56Traditional vs agentic automation
- 37:13Key takeaways
- 39:18Q&A: AI content
- 40:40Q&A: Prompt injection
- 41:46Q&A: AI automation tools
- 42:30Q&A: More AI opportunities
- 44:20Q&A: Developer involvement
related talks
24mScalable GenAI in Production
27mAI-Powered ETL Pipeline Orchestration: Multi-Agent Systems in the Era of Generative AI
16mPath to a More Artificially Intelligent Future
26mBuilding Agentic Workflows: Patterns for Orchestrating Intelligent Systems.
01h 23mBuild your own AI Agent
29mLeveraging AI to Enhance DevEx & Operational Efficiency
27mBuilding AI applications: Successes and Lessons
01h 07mNew Methods of Working with AI.