Skip to content

Back Matter

Acknowledgments

The Navy taught me that systems do not maintain themselves. Someone has to care enough to be awake at 3 AM with a flashlight and a checklist. That discipline transferred directly to database work, and that discipline is under everything Bob does.

Every DBA team I have worked with over twenty years contributed something to the diagnostic patterns in this book. The teams at Chickasaw Nation, TWIA, Lumeris, Entergy (through ComTec), and Dogsbarking each dealt with failure modes I had not encountered before, and each one is somewhere in the wait stats filter list or the system prompt templates. You taught me what to look for by making me look.

The AI-for-DBAs community is still early. The people who have engaged with the early versions of these ideas, challenged the prompt templates, pushed back on the auto-apply scope, and shared their own failure cases: you are the reason the prompt templates and the auto-apply scope have been stress-tested by more than one operator.

Get Involved

The work in this book is one DBA's home lab. The community is how it scales.

If reading this made you think I could run this against my own instances, there are three places to go from here.

Join the AI for DBAs community

The prompt templates, the failure cases, the pattern library, and the conversation about what to build next all live in two places:

Both are free. Both are where I post diagnostic recipes, system prompt revisions, and the war stories the book is too short to fit. Bring your own.

Send your war stories and ideas

A finding the agent missed. An extension you want to build. A wait-stats pattern that does not show up here. A configuration that makes Bob safer for your environment. Email me at [email protected].

I read every message. Failures are as useful as wins. The next version of this book will include the patterns that come back from readers actually running this in production.

If you are an investor

I am actively looking for backing to take Bob from a home lab project to a productized agent for managed-services providers and regulated-industry data teams. The architecture is proven. The stack is open. The compliance story is real. The market is every DBA who cannot send their query plans to a cloud API.

If you want to talk about funding the next stage, email [email protected] with subject line Bob investor inquiry. I will send the technical due-diligence package and we can find time to talk.

Bibliography and Source Materials

The technical claims in this book are grounded in production documentation that lives alongside the manuscript in source control:

  • Infrastructure truth: /home/ward/Projects/01-PRODUCTION/new-network/new-docs/NETWORK_DOCUMENTATION.md. Authoritative host inventory, IP assignments, open ports, and service dependencies as of the manuscript date.

  • Bob agent source: /home/ward/Projects/08-MY-AGENTS/Bob/docs. The production agent documentation including full system prompt history, probe query versions, and deployment runbooks.

  • SQL Server DMV references: Microsoft SQL Server documentation, sys.dm_os_wait_stats, sys.dm_exec_query_stats, sys.dm_exec_query_plan, sys.dm_db_index_physical_stats. All DMV queries in this book run against SQL Server 2019+.

  • MCP specification: Model Context Protocol specification, Anthropic, 2024. The MCP integration in Bob implements the tool-calling interface as specified.

  • Proxmox VE documentation: Proxmox VE 8.4.1 API reference. The snapshot and rollback calls in Chapter 8 use the REST API at /api2/json/.

Colophon

This manuscript was written in Markdown and version-controlled in Gitea at http://10.0.0.80:3000/hyp3rsoft/the-birth-of-bob.

Build toolchain:

  • Markdown source: one file per chapter, YAML front-matter per TestPlan.md ยง2
  • Diagrams: Mermaid, rendered to SVG by mermaid-cli during the build
  • Lint: markdownlint-cli and vale
  • PDF export: Pandoc 3.x with LaTeX
  • EPUB export: Pandoc 3.x
  • Static site: MkDocs Material with Glassmorphism CSS overrides
  • CI: Gitea Actions on gitea-runner at 10.0.0.80
  • AI assist for draft passes: local Ollama on inference-host-1 at 10.0.0.70, gemma4:26b for prose and reasoning, qwen2.5-14b on inference-host-2 at 10.0.0.71 for technical sections

The AI-density gate requires every chapter to score below 15% on ai-writing-patterns before release. The voice-check gate requires every chapter to match the ward-voice profile. Both run as part of the CI pipeline on every commit to main.

All code samples carry a // source: bob@c549c88 comment referencing the production Bob commit at the time of v0.1.0-rc1. Readers wanting the exact file each sample was adapted from can check out the Bob source at that commit and grep for the constant or function name shown in the snippet.

The Birth of Bob: CC BY-NC-SA 4.0 Code samples: MIT Ward T. Minson, 2026