Patrick Muggill
Enterprise systems for a living. AI experiments on the side. Mostly interested in where the two meet.
About
I'm a GTM Enterprise Architect at Splunk, a Cisco Company, where I advise leadership on architecture strategy, the global Salesforce footprint, and practical generative-AI adoption. Before Splunk, I spent four years at Accenture leading Salesforce delivery on large federal implementations, covering release management, DevOps, and custom development across Apex, Flows, and MuleSoft. Fifteen Salesforce certifications, and counting.
Most of my work today is about where generative AI meets enterprise reality: integration points, governance boundaries, failure modes, and the humans who end up owning the result. I spend more time on the seams than on the model itself.
I build smaller AI projects on the side. They sharpen the day job, and the day job sharpens them.
Outside of work: friends, heavy things, long runs, ancient history, and more hobbies than I can keep up with.
Now
Updated April 2026- Training for the Marine Corps Marathon (October 2026)
- Learning Go, the board game. Losing, mostly.
- Deep in Alexander the Great, and how one person reshapes a map
- Working through Brandon Sanderson's back catalog
- A couple of architecture notes I've owed myself for a while
Interests
- Enterprise AI that survives production
- Ancient history: Alexander, the Successor Wars, anything that rewires how you think about leadership and logistics
- Sci-fi and long-form fantasy: Brandon Sanderson, Andy Weir, Cixin Liu
- One Piece, a 25-year bet on long-form storytelling that's still paying off
- Weightlifting and pull-ups, the boring consistency of both
- DC sports, for better or worse
- Hosting friends, supporting local spots, chai lattes, sushi, and the occasional McChicken
- NMDP (Be The Match). I ran UVA's Be The Match on Campus chapter; 1,000+ donor registrations over three years, and still the easiest serious thing you'll do this year.
Projects
- Agentic retrieval sandbox An eval harness for comparing multi-step agent retrieval against plain RAG on messy, enterprise-shaped corpora.
- adr A small Go CLI for writing, linking, and searching architecture decision records across a repo.
- prod-ready-or-not A blunt checklist and linter for deciding whether an AI feature is actually ready for production.
Writing
- Apr 2026 Between the seams Why the hard part of enterprise AI lives between the boxes on the architecture diagram.
Goals
- Finish the Marine Corps Marathon, October 2026
- Reach Salesforce CTA (the long arc), for the parts of the job it forces you to master
- Play Go (the board game) well enough to lose gracefully to a 5-kyu
- Publish a handful of architecture notes I'd actually want to read
- Keep growing the bone marrow donor registry with NMDP
Contact
LinkedIn is the easiest way to reach me. I read messages.