Category archive

The architecture is what the product does when the model is wrong.

A model that is right most of the time is an architecture problem, not a prompt problem. These articles cover the surfaces that decide what the reader experiences on the other days: the contract on the output, the route to a second model, the trace that explains a bad answer, and the incident process for a system that is wrong fluently.

6 articlesProduct Architecture

Articles in Product Architecture

Product ArchitectureJun 30, 202613 min read
Incident Response for Systems That Can Be Wrong Fluently
A response playbook for quality regressions, prompt attacks, retrieval contamination, runaway agents, cost spikes, and provider failures.
Product ArchitectureFeb 26, 202610 min read
Observability for LLM and Agent Systems
Tracing model calls, retrieval, tools, state transitions, quality signals, and cost without turning telemetry into a privacy liability.
Product ArchitectureFeb 17, 202610 min read
Model Routing, Cascades, and Fallbacks
How to choose models per request using task risk, calibrated confidence, operational health, and total expected cost.
Product ArchitectureFeb 8, 202610 min read
Structured Outputs Beyond Valid JSON
Designing schemas, constrained decoding, semantic validators, repairs, and safe evolution for dependable model integrations.
Product ArchitectureDec 28, 202510 min read
AI Product UX for Systems That Can Be Wrong
Designing intent, progress, evidence, editing, confirmation, and recovery so capability becomes trustworthy product behavior.
Product ArchitectureDec 16, 202510 min read
Production AI Architecture in Laravel and PHP
Queues, streaming, typed provider boundaries, retrieval, tool execution, and observability for durable AI features in Laravel.

Apply the category

Put this against a real system.

If a decision in Product Architecture is in front of you right now, the fastest version of this is a call: bring the architecture, the failure you are seeing, and the constraint you cannot move.

Discuss the system

In the age of AI

The advantage was never the model. It's knowing what to build with it — and having a team that can actually ship it.

That's the part I help with: finding where AI genuinely makes your business faster, deciding what's worth building, and standing behind it once it's live.

Let's build it together See how we'd start

Four offices, one very full passport