Architecture for the implementation project layer across flagship tracks — milestone projects, capstone projects, operational exercises, and production-readiness criteria.
This document defines the project layer for all flagship tracks. Projects are not homework. They are operational work with production criteria — the same standards applied to real deployments.
The test for every project: Could A Square Solutions use this output to operate a real system?
If the project produces a document, system, or artifact that would be operationally useful in a real context — it's a real project. If it's an exercise designed to demonstrate understanding — it's homework. Homework doesn't belong here.
Project outputs are evidence. A completed project produces an artifact that the operator can:
Projects are not graded. The completion criteria are binary — either the artifact meets the specification or it doesn't. There's no partial credit rubric. The operator verifies their own completion against the published criteria.
Caps a module. Integrates all lessons in the module into one operational output.
Duration: 30–60 minutes
Output: One document, configuration, or procedure that could be deployed immediately
Verification: Checklist of production-readiness criteria
Caps a track. Integrates all modules into a comprehensive operational system.
Duration: 90–180 minutes
Output: A complete operational artifact: a system, playbook suite, or documented implementation
Verification: Multi-step production-readiness checklist + peer comparison criteria
Mid-module. Applies a specific lesson's technique to a real scenario.
Duration: 15–30 minutes
Output: A verified, working implementation of the lesson's technique
Verification: Specific functional test (not just "did you try it")
Module 1 (Foundations) — Milestone Project: Production Environment Audit
Output: A one-page Architecture Decision Record (ADR) for your AI engineering stack. Covers: tool selection with rationale, CLAUDE.md structure, permission configuration, version specifications, and the 3 failure patterns you've guarded against in your setup.
Production-readiness criteria:
.claude/settings.json has explicit allow/deny lists (not defaults)claude --version returns expected versionModule 2 (Prompt Engineering) — Milestone Project: Prompt Pattern Library
Output: A personal prompt pattern library — a reference document containing your 5 most common task types with their production-ready prompt templates, failure guards, and verification steps.
Production-readiness criteria:
Module 3 (GitHub Workflows) — Milestone Project: Git Recovery Playbook
Output: A personal git recovery playbook: the exact commands for your 3 most likely git failure scenarios with pre-conditions, recovery steps, and verification.
Production-readiness criteria:
Module 4 (Vercel Deployment) — Milestone Project: Deployment Safety Checklist
Output: A pre-deployment checklist for your specific project. Not a generic template — your actual deployment dependencies, env vars, build commands, and rollback plan.
Production-readiness criteria:
Module 5 (WordPress REST API) — Milestone Project: WP Automation Playbook
Output: A content operation playbook for one specific automation you run against your WordPress site. Documents: auth setup, the exact script/command sequence, dry-run output, verification procedure, rollback procedure.
Production-readiness criteria:
Track Capstone: Your Personal AI Operator Playbook
Output: A complete operational reference for your AI-assisted engineering practice. Includes: stack ADR, CLAUDE.md, top-5 prompt patterns, git playbook, deployment checklist, and documentation of one production operation you've completed.
Duration: 90–180 minutes to compile and verify
Production-readiness criteria: All 5 module milestone projects completed and integrated into one coherent reference document
Module 1 (Zero Budget Stack) — Milestone Project: Business Infrastructure Audit
Output: A verified infrastructure audit document for your actual AI business or planned project. Covers: current tool stack (with monthly costs), infrastructure architecture (GitHub + Vercel or WordPress), Analytics and GSC setup confirmation, and your 90-day traffic projection based on the organic traffic framework.
Production-readiness criteria:
This milestone replaces the final "review" lesson. If the infrastructure isn't set up for real, the module isn't complete.
Module 2 (First Product) — Milestone Project: Launched and Indexed
Output: Evidence that your first product or piece of content is live, indexed, and has received at least one organic visit.
Production-readiness criteria:
Track Capstone: AI Business Launch Dossier
Output: A complete launch dossier: infrastructure audit, product description, distribution channels with evidence, monetization status, and 30-day operating plan.
Duration: Compile from milestone outputs — 60 minutes
Module 1 (AI Search Mechanics) — Milestone Project: GEO Baseline Audit
Output: A baseline audit of one piece of existing content against GEO citation standards. Documents: entity density count per 500 words, answer-first compliance per H2 section, schema markup status, and Perplexity citation test result.
Production-readiness criteria:
Module 2 (Content Architecture) — Milestone Project: GEO-Optimized Content Piece
Output: One piece of content (minimum 1,000 words) written to GEO standards: answer-first structure, entity-dense, schema-marked, with verifiable Perplexity citation within 4 weeks of publication.
Production-readiness criteria:
Track Capstone: Full GEO Audit + 90-Day Strategy
Output: A complete GEO audit of one site with a 90-day implementation strategy. Includes: baseline citation rate, identified high-value content gaps, 12 specific articles to publish, and measurement plan.
Every project's criteria follow these principles:
1. Binary, not gradable. Each criterion is a yes/no check, not a quality assessment. "Analytics is configured" is valid. "Analytics is well-configured" is not.
2. Evidence-based. Where possible, criteria require a specific URL, screenshot, measurement, or output — not a self-report.
3. Failure-tested. Projects that involve deployment or automation must include evidence that a failure scenario was tested (dry-run was run, rollback was verified).
4. Date-stamped. Completed projects should include the date of completion in the artifact. Systems change; completion evidence needs to be dated.
Projects appear in tracks as type: 'project' lessons with status: 'available' once written. The lesson page for a project contains:
Progress tracking for projects works the same as lessons — CompleteButton marks completion locally. When auth is added, project completion requires evidence submission (artifact link or description).
Implementation project system v1.0 — 2026-05-18.