🚀 My documentation hub
Welcome to my experiments with using AI to retrieve, sanitize, validate, and create content. The underlying systems for these experiments are built to scale with requirements.
✍️ The modern Technical Writer
The core principles of technical writing remain unchanged: clear language, intuitive information architecture, and rigorous accuracy. However, the writer’s toolkit has expanded dramatically.
Instead of staring at a blank slate, writers now act as system architects who train, edit, and guide the AI:
- Eliminating the Blank Page: Leveraging LLMs for rapid first drafts and initial gap analysis.
- Designing for Agents: Structuring content so both engineers reading a webpage and AI agents executing an MCP query receive accurate, un-hallucinated context.
- Automated Guardrails: Using AI to review content at scale—detecting stale pages, flagging structural gaps, and syncing changes seamlessly (Create in GitHub → Publish to Confluence).
🏛️ What organizations need
To prevent AI hallucination and fragmentation, documentation infrastructure must fulfill core organizational requirements:
- Authenticity & Single Source of Truth: Duplicate or conflicting information confuses humans and AI agents. We need automated de-duplication and strict archiving of dated content.
- Consolidation of Disparate Sources: Pulling knowledge out of support tickets, customer calls, internal retrospectives, and code repositories into unified workflows.
- Internal Validation First: Before deploying conversational AI externally, we must test and validate RAG systems on internal datasets. If a bot wouldn’t satisfy your own team, don’t put customers through that experience.
🧭 Navigating the Portal
Use the left sidebar or the key entry points below to explore the documentation:
-
🕷️ Crawl and index
Automated crawling, indexing, and creating topics based on topics available on newsapi.org. -
🏗️ System Overview & Architecture
High-level system topology, deployment pipelines, and core service integrations. -
🛠️ Quality Guardrails Engine
Automated PR evaluation workflows powered by Gemini 2.5 Flash and GitHub Actions. -
💬 RAGbot Assistant Integration
Architecture and setup guide for the interactive site search assistant.
⚡ Quick Start
[!NOTE] This site has a lot of automation, including, but not limited to crawling, validating, and automatically creating PRs with missing information.