Skip to main content

Own System Design

5 min read

Cloud Arch

AI recommends patterns. You own cost, compliance, and 'what happens when AWS has a bad day.' That's the gap.

Data Arch

Schema and pipeline design are automatable. Data governance, semantics, and long-term evolution — that's your domain.

Solutions Arch

AI drafts solutions. You own client context, constraints, and the relationship. The proposal is a draft; the trust is yours.

Own System Design

TL;DR

  • AI can generate architecture diagrams and suggest patterns. It can't make the tradeoff decisions that fit your org, your constraints, and your risk appetite.
  • System design is your moat. Own it. Be the person who says "we should do X because Y" — and can back it up.
  • The architects who thrive are the ones who treat AI as a research assistant, not a replacement.

Yuki designs systems. AI can suggest architectures. The difference: Yuki knows the org's capacity, the legacy constraints, the political landmines. AI doesn't. That's the moat.

What AI Can and Can't Do

AI can: Propose patterns, generate diagrams, list tradeoffs from textbooks. AI can't: Know your org's budget, your team's skills, your compliance requirements, or the history of why the last "great idea" failed.

Your job is the gap. You take AI's suggestions and run them through the filter of reality.

How to Own It

  • Write the RFCs. Don't delegate the hard design work. Own it. Use AI for research and drafting; you for the actual decisions and rationale.
  • Build institutional memory. When you make a decision, document why. "We chose X over Y because Z." That becomes the reference. You become the authority.
  • Be the tie-breaker. When teams disagree on architecture, you're the one with the deep context. Use it. Make the call.

The "AI suggested it" Trap

Don't hide behind AI. "Claude said we should do microservices" is not an architecture decision. "We're doing microservices because [org-specific reasons], and here's the migration plan" is. Own the reasoning. Own the consequences.

Quick Check

AI suggests a microservices architecture. It's technically sound. As a Staff engineer, what's the right move?

AI generates an architecture diagram. You present it. 'Claude said we should do this.' Someone asks why. You don't have org-specific reasons. You look like you're hiding behind the tool.

Click "I decided because" to see the difference →

Do This Next

  1. Identify one system you could own more fully — the one where you're already the go-to. Formalize it: write an ADR, update the runbook, make it explicit.
  2. Use AI for your next design — but document your overrides. "AI suggested X; I chose Y because Z." That's the audit trail that proves you're thinking.