Skip to content

Perspectives and analysis.

Articles and analysis on enterprise architecture, AI engineering, and delivery discipline.

CMX in 2026: what exists today

A governed AI gateway in three languages, a workflow engine with measured numbers, a method that attacks our own designs, and twelve chairs for interns. What we built this year, and what is still missing.

We checked our own design three ways. Ten problems came out.

Two nights designing a truck tracker. Version 1 looked serious. A state exploration, a numeric check and a resilience analysis disagreed, and version 2 is a different product.

We killed our workflow engine mid-transaction. On purpose.

hwf is our C++20 prototype of a durable engine for processes that mix code, AI agents, people and timers. What it guarantees, what we measured, and what it still lacks.

Eleven controls before a single byte reaches a model provider

Central IA Souveraine is the governed door we built for regulated institutions: one entry point, eleven controls, and a ledger an auditor can verify. What it does, and what it costs to run.

CMX does not just architect. We solve.

Architecture without delivery is theory. CMX operates at every level - from strategic clarity to production systems - because our clients need results, not diagrams.

Agentic AI needs governance before autonomy

Autonomous agents are powerful. Without guardrails, decision traceability, and human-in-the-loop validation, they become liabilities.

Why AI without architecture produces chaos

Why AI without architecture produces chaos

Generative AI multiplies initiatives. Without clear architecture, it also multiplies technical debt and inconsistencies.

Integration debt is the silent killer of enterprise velocity

Most enterprises don't have a technology problem. They have an integration problem that compounds silently until everything slows down.

Enterprise architecture is not a diagram. It is a decision system.

EA has a reputation problem. Most frameworks produce documentation nobody reads. The real value is in decision quality, not artifact quantity.

From AI proof-of-concept to production: the industrialization gap

Most AI initiatives succeed in the lab and fail in production. The gap is not technical - it is structural, organizational, and architectural.

Designing systems that absorb change instead of breaking

The question is not whether your system will face change. It is whether your architecture can absorb that change without a rewrite.

Let's structure your next intelligent system.

Whether you are launching an AI initiative, modernizing architecture, or aligning technology with execution, CMX helps you create the structure required to move with confidence.