The Brief History of Artificial Intelligence
Chapter 8: Methodology

Production

2 min read

Production Workflow: From Research to Publication

The project was completed in 5 intensive days (5-6 hours of focused work per day), following a workflow structured in three major phases.

Phase 1: Research and Documentary Consolidation (Days 1-2)

- Research via MCP (connections to academic sources)

- Indexing facts by historical period

- Creating a structured documentary base

- Validating primary sources

Tools used:

- MCP to connect academic databases

- Indexing in VS Code to reference sources

- Verification agents to cross-check information

Phase 2: Structured Writing with "Single Source, Three Outputs"

The principle: Each written article serves as a single source for three different formats.

From a single written article (~1,500 words) with three-part structure, extended metaphor, and scientific rigor, three outputs are generated:

- Podcast (Spotify, Apple)

- Video (YouTube, future)

- Web (history-ai.us)

Why this approach?

- Ensures narrative consistency across formats

- Optimizes production time

- Allows for format-specific adaptations

Phase 3: Publication and Technical Deployment (Days 4-5)

- Web integration (Vercel + Next.js): 40+ pages generated with structured content

- Podcast publication:

- Audio generation (ElevenLabs)

- Chaptering and metadata

- Going live (Spotify <10min, Apple Podcasts)

- Multi-platform promotion strategy:

- Reddit (tech + history communities)

- LinkedIn (professionals)

- WhatsApp (targeted audiences)

Tools and Technologies

Content Production Stack

ToolFunction
VS Code + Claude CodeAugmented IDE enabling management of 40+ Markdown files simultaneously
MCP (Model Context Protocol)Integration of external documentary databases
Custom skillsEditorial automation (avalon-author, verification agents)
External APIsConnections for validating dates, historical facts

Publication Stack

ToolFunction
VercelWeb hosting with instant deployment and performance
Next.jsWeb framework with static generation and optimized SEO
ElevenLabsAudio generation with voice quality and studio interface for chaptering
Spotify for CreatorsPodcast distribution with publication <10 minutes
Apple PodcastsInternational podcast audience

Technical Note: Image Generation

Identified limitation: Image generation in VS Code/Claude Code is not as mature as text generation.

Adopted solution:

- Deliberately simple visual style to facilitate generation

- Short, repeatable visual prompts

- Fewer variables than for text