Ideas are cheap. Systems are leverage.
This is the execution stack: tools that turn plain English into working software and automated workflows. Prompt → app → workflow → revenue. Start here.
How to use this page
Three steps. Pick the right tool, follow the playbook, copy the prompt. In that order.
Pick your tool by job
Writing real code? Use Cursor. Building an MVP from scratch? Use Lovable. Automating across SaaS? Zapier or n8n. Match tool to job — not to hype.
Run the right playbook
Prompt → MVP, Manual → Automated, or Agent with guardrails. Each is a proven sequence — don’t skip steps or mix patterns until it’s working.
Copy the prompts
Paste-ready prompts for MVP specs and automation design are built in. Use them to define scope before you generate — it cuts rework in half.
Tool map — pick by job
Four tools. Four specific jobs. Pick the one that matches what you’re building right now.
Write and refactor real code inside your editor. Scaffold fast, tighten logic, stay in control of what ships.
- Best move: scaffold fast, then tighten logic with tests
- Risk: blindly accepting code you don’t understand
Turn prompts into full-stack MVPs. Define pages, data, and actions first — then generate. Ship before you add features.
- Best move: define pages + data + actions before generating
- Risk: scope creep — ship a small MVP before “features”
Automate across thousands of SaaS tools. Start with simple triggers, then layer in agent steps — with approval gates on anything consequential.
- Best move: simple triggers first → agent steps later
- Risk: unbounded agents — keep approvals on money/actions
Self-hosted automations with full data sovereignty. Build stable workflows first, then add AI steps. You own uptime — plan for it.
- Best move: stable workflow first, then AI steps
- Risk: maintenance burden — self-hosting means you own uptime
3 playbooks that ship — not theory
These patterns turn “cool tools” into repeatable outcomes. Pick one and run it end to end before mixing patterns.
Prompt → MVP
Define user → define pages → define data → generate → ship. Use Lovable to build; use Cursor to clean. Goal: working app in 1–2 sessions.
Manual → Automated
Map steps → set trigger → transform data → notify → log. Start with Zapier. Move to n8n when you need data control. Goal: remove recurring busywork.
Agent with guardrails
Define allowed actions → define approvals → add logs → add rollback. Humans approve; agents execute; logs prove. Goal: safe “AI teammate” behavior.
Copy prompts — builder grade
Use these to produce clean requirements before you generate. Scoping first cuts rework by half.
MODE: BUILD (MVP)
TASK:
Turn this idea into a simple MVP spec.
IDEA:
[Describe the app in 2–4 sentences]
OUTPUT:
1) User types (1–2)
2) Pages (max 5) with purpose
3) Data objects with fields
4) Actions (create/read/update/delete)
5) One must-have workflow automation
6) First version constraints (what NOT to build)
MODE: AUTOMATION DESIGN
TASK:
Design a workflow automation.
INPUTS:
- Trigger:
- Source system:
- Destination system:
- Data needed:
- Rules/conditions:
- Human approval points:
OUTPUT:
- Step-by-step workflow
- Failure handling
- Logging requirements
- Security notes (PII, payments, comms)
Next up: Data, RAG & Discovery
The final page in this stack explains how AI remembers, retrieves, and cites your data — so it stops guessing and starts proving.
Governed by: AI Bill of Rights • AI Constitution • Monetize this stack: AI Blueprint™ Business