SA-101 · Module 3

The 5-Minute Architecture

3 min read

I measure solution health by what I call "mean time to explain." If I cannot explain the architecture to a non-technical stakeholder in five minutes, the architecture is too complex. Not the explanation — the architecture. Complexity that cannot be communicated is complexity that cannot be maintained, and complexity that cannot be maintained will eventually fail.

The 5-minute architecture is not a summary of the full design. It is the structural narrative — the story of how the system works, told at the level where business stakeholders can follow the logic and technical stakeholders can see the skeleton. It answers three questions: what goes in, what happens inside, and what comes out. Everything else is detail that the full architecture document covers.

  1. Minute 1: The Problem State the business problem the architecture solves. One sentence. "Your team spends 8 hours a day manually categorizing support tickets that an AI system could triage in seconds." This anchors the entire explanation in something the stakeholder cares about.
  2. Minute 2: The Inputs What data enters the system and from where? "Tickets arrive via your existing Zendesk integration. The system also ingests your product documentation and past resolution history." The stakeholder now understands what feeds the machine.
  3. Minute 3: The Processing What does the system do with the inputs? One to two sentences, no jargon. "The system matches each new ticket against your documentation and past resolutions to suggest the most likely category and a draft response." This is the architecture in a single breath.
  4. Minute 4: The Outputs What does the stakeholder see? "Your support team sees a pre-categorized ticket with a suggested response they can approve, edit, or override. The system learns from their corrections." Concrete, visible, and directly tied to the business outcome from minute one.
  5. Minute 5: The Safeguards What protects the stakeholder? "A human always reviews before a response is sent. The system logs every decision for audit. If confidence is below 80%, it escalates to a senior agent instead of suggesting a response." This is where trust is built.