AI Engineering That Ships

Beyond Code Completion: Building an AI Development Methodology with GitHub Copilot

We asked Copilot to add a customer credit hold to our billing module. It wrote syntactically correct code that would have silently corrupted the data isolation that the entire application depends on. The logic looked plausible. The business rule it violated is not documented anywhere. That failure forced a question: what does it actually take for AI to work in a real enterprise codebase? Figure 1 - The Enterprise AI Gap: Code completion operates on syntax. Enterprise business logic operates on decades of accumulated rules that exist only in running code and specialists' heads. Bridging that gap requires a different approach entirely. Most…

Beyond Code Completion: Building an AI Development Methodology with GitHub Copilot

From YouTube to Knowledge Graph: Building an AI-Powered Content Pipeline

1,000+ videos. 2,757 auto-generated links. $1.50 in API costs. Here's how we built it. By the Dotzlaw Team The Achievement We turned 1,000+ YouTube videos into structured, interconnected Obsidian notes -- complete with YAML frontmatter, key takeaways, timestamped sections, and 2,757 bidirectional links that wire the entire vault into a navigable knowledge graph. The total API cost was $1.50. The total processing time was 30 minutes. Every note connects to 3-5 semantically similar notes, every tag maps to a curated taxonomy, and the whole thing runs from a single web interface where you paste a URL and get back a finished note. This…

From YouTube to Knowledge Graph: Building an AI-Powered Content Pipeline

Claude Code

An Agent Swarm That Builds Agent Swarms: How We Used Claude Code to Generate Claude Code Infrastructure

We used Claude Code to build a framework that generates Claude Code infrastructure for any project. Then we proved it works by migrating two production apps -- the second more complex but completed faster. 67 Python files, an AI/ML pipeline, a full architectural redesign, 8 sessions, zero framework build time. Figure 1 - The Three-Folder Architecture: The framework reads the source project but never modifies it. All generated infrastructure lands in a fresh target project. This READ-ONLY invariant held across 10 sessions and was never violated. A user types "top 10 customers in revenue for the last 3 years" into a text box. Thirty seconds…

An Agent Swarm That Builds Agent Swarms: How We Used Claude Code to Generate Claude Code Infrastructure

AI Security

Adversarial Agent Testing: When Your AI Agents Attack Each Other

Five Claude Code agents. Three teams. One target. The Red Team found 7 vulnerabilities in 5 minutes. The Blue Team patched every one. Then Round 2 proved that patches alone aren't enough. Figure 1 - The Adversarial Architecture: Red Team (recon + exploit agents) attacks the target from one worktree. Blue Team (monitor + hardener agents) defends from another. The Referee scores both sides from a read-only vantage point. Information asymmetry is enforced by hooks -- not prompts. Two rounds of adversarial exercises against a real application. 27 total attacks across 10 OWASP categories. 14 defense patches adding 550 lines of security…

Adversarial Agent Testing: When Your AI Agents Attack Each Other

Claude Code

Orchestrating AI Agent Teams: How Skills, Hooks, and Context Flow Make Autonomous Coding Reliable

An orchestrator breaks a task into pieces. Specialized agents pick up work items, each carrying skills that define what they know and hooks that enforce how they behave. Context flows from session start to task completion through a deterministic pipeline. Here is how the pieces fit together. Cover - The Orchestrator Pattern: A central orchestrator coordinates specialized agents, each equipped with domain-specific skills and hooks. Context flows bidirectionally: the orchestrator assigns tasks with context, agents execute with skill-guided knowledge and hook-enforced quality, and results flow back through the same deterministic…

Backend Development & Data Analytics

Orchestrating AI Agent Teams: How Skills, Hooks, and Context Flow Make Autonomous Coding Reliable

The Dotzlaw Team

Two skilled engineers building advanced agentic AI projects and research alongside me. They contribute directly to the systems, articles, and tools published on this site.

Katrina Dotzlaw

katrina.dotzlaw.com ↗

Building AI-powered data pipelines and full-stack applications at the intersection of machine learning and real-world business problems.

Python Java Machine Learning UI Design

Visit Site →

Ryan Dotzlaw

ryan.dotzlaw.com ↗

Software Developer & Data Analyst

Applying statistical analysis, neural networks, and modern UI to extract insight from complex datasets and build compelling data-driven applications.

Python R Neural Networks UI Graphics

Visit Site →

Latest Insights

View all →

Part 5 of 5 Copilot Agent Pipelines

What Building an AI Development Methodology Taught Us About Enterprise Software

Five lessons from building 7 specialized Copilot agents, a Neo4j code graph indexing 10,000+ functions, and a self-improving knowledge system with 18 domain skills for a large-scale enterprise codebase. The gap between AI demos and enterprise reality is not technology. It is methodology.

2026-03-26 Read Article →

Part 4 of 5 Copilot Agent Pipelines

Self-Improving AI: How Code Reviews Feed a Knowledge Flywheel

Every code review harvests knowledge. Knowledge updates skills. Better skills produce better code. Eighteen domain skills and growing, each one making every Copilot agent smarter in that domain. Here is how we built a system that gets better every time someone uses it.

2026-03-25 Read Article →

Part 3 of 5 Copilot Agent Pipelines

Neo4j Code Graph: How Graph-Based Code Intelligence Changes What AI Agents Can Do

Text search finds where a function name appears. A code graph tells you who calls it, what it calls, and the full call tree from any entry point. We indexed 10,000+ functions into Neo4j and built agents that query it directly. The first pilot mapped 21 functions across 11 tables in 30 minutes.

2026-03-24 Read Article →

Part 2 of 5 Copilot Agent Pipelines

The Development Workflow: How Seven Agents Turn a Ticket into Reviewed Code

One AI agent cannot research, plan, implement, review, and document effectively. Seven specialized agents can. Here is how we built a structured development workflow with handoff buttons, file-based artifacts, and cross-model orchestration for a large-scale enterprise codebase.

2026-03-23 Read Article →

Part 1 of 5 Copilot Agent Pipelines

Beyond Code Completion: Building an AI Development Methodology with GitHub Copilot

GitHub Copilot suggests a line of code. Our enterprise codebase has 10,000+ functions across 22 modules. The gap between code completion and business context is where most AI adoption stalls. We closed it with 7 specialized agents, a code graph database, and a self-improving knowledge loop.

2026-03-22 Read Article →

Part 5 of 5 Obsidian Notes Pipeline

Ask Your Vault Anything: Building a RAG Chatbot for Your Obsidian Notes

A RAG chatbot that answers questions about your Obsidian vault in 2.5 seconds with source attribution and one-click navigation to source notes.

2026-03-14 Read Article →

Part 4 of 5 Obsidian Notes Pipeline

Obsidian Vault Curation at Scale: How We Transformed 1,000+ Notes in Under an Hour

1,280 chaotic tags, three different frontmatter formats, fixed in 30 minutes for $1.50 using AI-powered batch processing.

2026-03-13 Read Article →

Part 3 of 5 Obsidian Notes Pipeline

Building a Semantic Note Network: How Vector Search Turns Isolated Notes into a Knowledge Graph

1,024 notes, zero manual links, 2,757 bidirectional connections discovered automatically using vector search and semantic similarity.

2026-03-12 Read Article →

Part 2 of 5 Obsidian Notes Pipeline

Anthropic Batch API in Production: 50% Cost Reduction Through Smart API Architecture

782 files, 8 batches, 25 minutes. Building a dual-mode API architecture that automatically chooses between real-time and batch processing for 50% cost savings.

2026-03-11 Read Article →

Production Projects

FastAPI · React 18 · PostgreSQL

Obsidian Notes Pipeline: AI-Powered Knowledge Management

A full-stack RAG application that transforms YouTube videos into interconnected Obsidian notes -- 1,000+ notes, 2,757 auto-generated links, 5,000 searchable chunks, and a chatbot with 2.5s latency, all for $1.50. 1,000+ notes → knowledge graph · 2,757 bidirectional links · 2.5s RAG chatbot response · $1.50 total pipeline cost

Claude Code · Python · UV Scripts (PEP 723)

Obsidian Notes Pipeline: AI-Powered Knowledge Management

Claude Code Bootstrap Framework

An agent swarm that builds agent swarms. A 12-step pipeline where Claude Code agents analyze any codebase and generate complete Claude Code infrastructure -- agent teams, hooks, skills, and slash commands -- in 30-55 minutes. Three production migrations validated. The second was harder but faster. 3 production migrations validated · 17 skills + 17 hook templates · 12-step generation pipeline · 10/10 OWASP coverage

Claude Code · Python · UV Scripts (PEP 723)

Adversarial Agent Testing

AI agents that attack each other to find vulnerabilities. Red Team probes, Blue Team defends, a Referee scores both -- all using Claude Code with worktree isolation. Two rounds of live exercises against a real target drove ASR from 65% CRITICAL to 47% HIGH, with a regression wave proving patches hold at 20% and an escalation wave exposing architectural gaps at 85.7%. ASR 65% → 47% across 2 rounds · 27 attacks, 14 defense patches · 10/10 OWASP category coverage · 5 specialized agents

GitHub Copilot · Neo4j · VS Code

GitHub Copilot Agent Pipelines

Seven specialized Copilot agents that form a structured development workflow for a legacy Servoy enterprise application with 10,000+ functions, 1,000+ files, and 22 modules. Neo4j graph-powered code intelligence, cross-model orchestration, and a self-improving knowledge loop where every code review makes every agent smarter. 7 specialized agents in a structured workflow · 10,000+ functions indexed in Neo4j graph · 18 domain skills with self-improvement · Cross-model orchestration (4 models)