§ 03 · Writing

Notes from systems that had to keep running.

I republish short LinkedIn posts here when they deserve to live somewhere more stable. The point is not thought leadership. The point is to leave a proof archive: decisions, constraints, and failure modes from production AI work.

§ 01

Recent articles

Republished from production notes

012026 · 04 · 6 min · DraftThe strict-schema discipline, and what it costs youA short note on why schema locks make AI systems less magical and more useful.Schema lockReliabilityProduction AI 022026 · 03 · 4 min · DraftScrape, integrate, or ask: a decision sketchHow I decide whether a workflow needs a browser, an API, or a human in the loop.ArchitectureAutomation 032026 · 02 · 9 min · DraftSix-source synthesis - notes from productionThe model is rarely the hard part. Reconciliation is where the system earns trust.GTM intelligenceResearch systems 042026 · 01 · 5 min · DraftOn being the only engineer on the projectSolo build work is mostly about deciding what the system should refuse to do.Solo buildFractional builds 052025 · 11 · 7 min · DraftPlaywright in production: the boring infrastructureCookie refresh, observability, retries, and the parts of browser automation nobody screenshots.PlaywrightOperations 062025 · 10 · 5 min · DraftWhat I owe a system after handoffThe work is not done when the code is deployed. The operational contract matters.OwnershipMaintenance

Have a system like this to ship?

For focused AI systems architecture or production workflow work, write directly. I take on limited, non-conflicting fractional builds.

hello@chrisgarre.com

Directhello@chrisgarre.com

Cadence2 business days

EngagementsFractional · 6-12 wk

CapacitySelective · 1-2 d/wk