m@berryhill: ~/berryhill.dev/about.md
~ homeposts/about.md
m@berryhill in ~ $cat about.md
---
name: Matthew Berryhill
role: engineer + writer + operator
based: Los Santos, Panamá
nationality: USA
writing: field notes on AI-native systems
focus: [agent workflows, governance, provenance, review gates]
contact: matt@berryhill.dev
---
// $ whoami --verbose

I build AI-native systems and write about the operator work behind them.

This site is about agent workflows, governance, provenance, memory, handoffs, escalation, and the review discipline required to turn autonomous output into shipped proof. The background is not a minimalist placeholder: it runs through software engineering, DevOps, product systems, digital music, creative businesses, and the agentic-first work that now anchors berryhill.dev.

Nationality
USA
American operator
Location
Los Santos, Panamá
remote operator
Systems focus
agents · review · provenance
operator loops over hype cycles
Experience arc
2010 → present
music, startups, telecom, product, agents

What I actually do

The honest answer is: I build and inspect AI-native operating loops.

Engineering. I work on the code, review gates, publishing surfaces, and glue that make agent work auditable instead of magical.

Writing. I use essays as field notes: what broke, what shipped, what needed a human decision, and where the agent loop produced real leverage.

Strategy. I care about capital markets, crypto rails, AI/ML systems, and automation only when the discussion gets specific enough to change a build decision.

Things I believe

Not a manifesto. A fast filter for the kind of work this site takes seriously.

01
Agentic development is real, but the durable work is governance. The useful question is not whether an agent can produce output. It is who reviews it, what persists, what escalates, and how the next run gets better.
02
Provenance beats vibes. If a workflow cannot show its sources, handoffs, assumptions, and review trail, it is not ready to carry serious decisions.
03
The operator loop is the product. Tools matter, but the advantage comes from disciplined decomposition, verification, memory, and escalation.
04
Specificity is the cheapest leverage. Name the constraint, the artifact, the gate, and the failure mode. Otherwise the copy is probably hiding weak thinking.

How I got here

A practical timeline: digital music and creative systems, contract engineering, telecom and product engineering, founder/operator work, then agentic-first building and writing at berryhill.dev.

Oct 2025 – Present
Builder
berryhill.dev · Self-employed · Remote
Personal brand representing Matt as a digital builder focused on AI-native systems, agent workflows, governance, provenance, review gates, handoffs, memory, escalation, and shipped proof.
Sep 2023 – Sep 2025
Contract Developer
Tronic · Full-time · Remote
Senior development roles.
Oct 2019 – Sep 2023
Founder / CEO
Pineapple Workshop · Self-employed · Denver, Colorado, United States
Founder/operator work at the intersection of software, creative systems, and product execution.
Aug 2018 – Oct 2019
Software Engineer III
Comcast · Denver Metropolitan Area
Software Engineer III via Turnberry Solutions. Product Engineering for Commercial Voice.
Apr 2018 – Jul 2018
Software Engineer
Charter Communications · Denver Metropolitan Area
Software Engineer at Charter Communications via Apex Systems. Video Operations.
Oct 2017 – May 2018
Contract DevOps / Software Engineer
Effortless Rental Group · Denver Metropolitan Area
Contract DevOps and software engineering work.
Jan 2017 – Aug 2017
Software Engineer
Beatport · Denver Metropolitan Area
Software engineering in digital music infrastructure.
Mar 2016 – Jan 2017
Contract Developer
MachineShop · Denver Metropolitan Area
Contract development work.
2010 – 2015
Founder/Musician
Robotic Pirate Monkey LLC · Boulder, CO
Founder and musician of the electronic music trio Robotic Pirate Monkey.

How I work with people

The shape of serious work is roughly the same:

  • Start with the real constraint, not the deck constraint.
  • Map what needs to be built, proven, measured, or explained.
  • Ship small artifacts that make the next decision less fake.
  • Keep the operator loop visible.

FAQ

Are you available?

Email the specific problem, artifact, and decision you need to make. I am selective about calls and avoid vague AI strategy theatre.

What should I send first?

Send the problem, the deadline, the current artifact, and the decision you need to make.

What do you not do?

I do not do token launches without real rails, generic AI punditry, or content that hides the underlying work.

Let's connect..

Skip the vague AI strategy call. Send the problem, artifact, deadline, and decision you need to make.

← cd ../ back to ~EOF · field notes · © Berryhill 2026 · rss