Module 4 The Lean SAE Tech Stack

Module 4: The Lean SAE Tech Stack

Status: Complete · The Zero‑Ops principle and a reference stack. [Contributions Welcome ✍️]

---

4.1 Cognitive load minimization

Every technology in an architecture carries a long‑term maintenance tax. If a framework, cloud provider, or database engine requires a dedicated specialist to deploy, scale, configure, or tune it, it is banned under the SAE framework. The target stack is Zero‑Ops: provisioning, scaling, and routing are handled by the platform, not by you.

The test for any tool:

Can a single operator adopt this, ship to production, and stay in production — without becoming its part‑time administrator?

If no, it's friction.

---

4.2 The "Zero‑Ops" core reference stack

Tier	Legacy corporate choice	SAE selection	Friction eliminated
Workspace & Execution	Local IDE + Jira + Slack + standups	Cursor + Claude Code	Ticketing backlogs, manual syntax, status‑alignment updates
Frontend Foundation	Webpack + custom React configs	Next.js + v0 + Shadcn	Build‑pipeline setup, UI boilerplate, manual CSS debugging
Data & Persistence	Self‑hosted RDBs (AWS RDS / Oracle)	Supabase / Neon Postgres	DB provisioning, connection‑pool tuning, API route scaffolding
Infrastructure Compute	Cloud VMs (EC2) + Kubernetes + Docker	Vercel / Cloudflare Edge	Server provisioning, container mgmt, reverse proxies, multi‑region routing

This is a reference stack, not a mandate. The principle (Zero‑Ops) is the law; the specific tools are interchangeable. Submit alternatives to the Approved Zero‑Ops Tooling List.

---

4.3 Choosing a tool: the Zero‑Ops scorecard

Rate a candidate tool 0–2 on each axis. 8+ = SAE‑grade.

Axis	0	1	2
Provisioning	Manual servers	Some config	Instant / managed
Scaling	Manual capacity planning	Autoscale w/ tuning	Invisible / automatic
Ops burden	Needs a dedicated admin	Occasional intervention	Effectively zero
AI‑friendliness	Opaque, hard to generate against	Workable	First‑class SDK + clear docs models can use
Exit cost	Deep lock‑in, no export	Some lock‑in	Standard formats, portable

Reality check. Zero‑Ops trades operational control for vendor dependence. That's usually the right trade for a solo operator — but go in with eyes open: managed platforms can have surprise pricing at scale, regional/compliance gaps, and migration cost if you outgrow them. Favour tools with standard exit formats (e.g. plain Postgres you can pg_dump) so "Zero‑Ops" never becomes "zero leverage." For regulated data‑residency requirements, verify the platform's region story before committing — see When NOT to Use SAE.

---

▶ Putting it into practice: the Approved Zero‑Ops Tooling List (scored picks) · lock‑in and cost trade‑offs in Risks, Security & Governance.

Next: Module 5: Project Navigation & Metrics →