Three AI Providers Went Down at Once. Would Your Product Keep Working?
The September outage cluster exposed a simple production risk: an AI feature needs a tested degraded mode, not just a second model endpoint.
K
Kyluke McDougall Insights
Insights
Thoughts on software, business, and building things that matter
Topic guides
Guidance organised around your technical decision
01
Software modernisation
Legacy systems, make-or-buy decisions, and technical renewal without a reckless rewrite.
Open guide and articles 02Apps and digital products
From product judgement and prototypes to dependable apps, platforms, and SaaS products.
Open guide and articles 03Software architecture and operations
System boundaries, backends, cloud infrastructure, and decisions that keep software maintainable.
Open guide and articles 04Production-grade AI agents
Security boundaries, governance, and delivery practices for AI agents in real systems.
Open guide and articles 
DeepSeek V4 Flash Is Codex-Compatible. That Does Not Make It Drop-In Safe.
DeepSeek's new Responses API support makes model evaluation easier, but serious teams still need compatibility tests, security boundaries, compliance checks, and a rollback plan.
K
Kyluke McDougall 
An AI Agent Reached Production. Your Sandbox Is Not the Boundary.
The OpenAI and Hugging Face incident turns agent containment from a theoretical concern into an architecture and operations requirement.
K
Kyluke McDougall 
MCP Went Stateless. Your Agent Architecture Still Has State.
MCP 2026-07-28 removes protocol sessions and makes remote servers easier to scale. Here is what production teams should redesign before upgrading.
K
Kyluke McDougall 
AI coding agents need security gates before they touch production
Agentic coding tools can now change files, run commands, and influence delivery. Serious teams need production gates for secrets, dependencies, tests, architecture, and auditability.
K
Kyluke McDougall 
Loop Engineering: Why AI Coding Agents Need Architecture, Not Just Prompts
AI coding agents become useful in production when teams design feedback loops around them: tests, review, security gates, telemetry, rollout control, and clear ownership.
K
Kyluke McDougall 
AI Coding Needs Quality Gates, Not Vibes
AI coding tools can accelerate delivery, but only if teams measure quality, reviewability, security, test evidence, and traceability instead of trusting plausible output.
K
Kyluke McDougall 
Zero Trust for AI Agents: The Architecture Work Before Autonomy
AI coding agents can speed up delivery, but production teams need bounded permissions, short-lived credentials, policy gates, sandboxing and audit logs before giving agents real access.
K
Kyluke McDougall 
Secure Agentic Coding Needs Governance, Not Just Better Tools
AI coding agents are becoming part of software delivery. The hard question for product teams is how much autonomy they get, what they can access, and how their work is verified before it reaches production.
K
Kyluke McDougall