Capstone: Automate a Real Workflow
Build, measure, and hand off a working automation for a real San Antonio business or community group.
Objective
By the end of this course, you hand your partner a working automation, honest proof of what it changed, and the documents they need to run it without you.
What you have built so far
Look back. You already have most of the capstone:
- Week 1: a shortlist of tasks to automate, assist, or abstain
- Week 2: a workflow map and a baseline
- Week 3: a working intake automation
- Week 4: AI steps with human approval
- Week 5: clean data, error alerts, and a break plan
- Week 6: a privacy check and a handoff draft
- Week 7: after numbers and a tracker
The capstone pulls it together for one real partner.
Capstone
Capstone brief
- A San Antonio small business, nonprofit, church office, or youth program you recruited in Week 1
- Your partner loses hours each week to one repeat workflow. Automate it end to end in Zapier, Make, or n8n, keep people in charge of the steps that need judgment, and prove what changed with before and after numbers.
Deliverables
- A working automation running in an account your partner owns
- Your workflow map, before and after, with every human step marked
- A one-page automate, assist, abstain decision record that says what you chose not to automate and why
- Baseline and after measurements, with a time-saved calculation that includes upkeep
- A privacy check showing what data flows where and what never enters an AI step
- A handoff document with the break plan and how to turn it off
- A tracker tab your partner can read in under a minute
- A 3-minute demo video with captions, or a live demo at the final Cook Up
| Criterion | Meets the standard when |
|---|---|
| Judgment | Explains what was not automated and why. Human approval sits on every step that reaches the public. |
| It works | Runs on real or approved data for at least one week. Normal, missing, and odd inputs are handled. |
| Data hygiene | Fields are cleaned, there is one source of truth, and errors alert a named person. |
| Privacy and safety | Least data, least access. No minors' data in AI steps. Written permission from the partner. |
| Honest measurement | Before and after measured the same way. Upkeep counted. Claims match the data. |
| Handoff | Partner owns the accounts and can run, pause, and get help without the builder. |
| Clarity | Documents use plain language. The video has captions. A new staff member could follow them. |
Finding a partner: ask a business you already go to, a nonprofit where you volunteer, your church office, or a program at your school. Coaches can connect you with a Community Incubator partner site if you need one. If no partner works out, build for a realistic made-up organization using fake data, and say so clearly in your demo.
Example capstones
These show the size we are aiming for. One workflow, done well.
- Taquería catering intake: form and Facebook requests flow to one sheet, AI summarizes each one, AI drafts a reply saved to Gmail for the owner to send.
- Food pantry volunteer shifts: sign-ups log to one sheet, reminders go out two days before, no-shows are flagged for the coordinator. Family requests stay fully human, with an alert to staff.
- Barbershop reminders: bookings trigger a reminder text the night before with a reply link to reschedule. The barber keeps his own judgment on fitting in walk-ins.
- Youth league game days: a weekly template message goes to parents and guardians with time and gym address. No youth data leaves the league's sheet.
Final check
Before I submit my capstone
0 of 7 doneQuiz: Module 7 check: measure and ship
Go deeper: Automated Workflows: The Playbook for Cutting Busywork by Dom Davis, Temerarii Media.