The Operator OS Module Map
Every module in the stack that runs 7 donut shops — what each one replaced, what it cost to build, and how they connect.
Preview the artifact →13 years in tech. October 2024 I planned my way to a layoff and walked out with severance. 18 months later: two physical-asset businesses, 20+ employees on the kitchen floor, custom software replacing the third-party stack — donut shop ops and property management both.
13 years in tech. 6 companies. Six-figure salary, stock vesting. Every Sunday night I dreaded Monday and told myself it was just a season. I told myself for five years.
Then I made a decision. I didn't quit. I planned my way to a layoff in October 2024 and walked out with six-figure severance. By that point I had 5 STRs cashflowing and 12 months of savings stacked. Tech was deep in layoff cycles. I let the timing work.
Eighteen months later: 11 STR/MTR units across Hawaii, DFW, and Tampa. 7 donut shops in DFW. The first ops stack took 3 weeks because the off-the-shelf tools cost too much and still didn't fit a 4am deskless workforce.
Donut shops, insurance housing, and the internal tools built because the third-party stack didn't fit.
7 donut shops across DFW. 20+ employees, most of them not sitting at laptops. The first problem wasn't strategy — it was getting daily checkins, schedules, inventory, cash counts, and investor updates out of tools built for desk workers.
11 STR/MTR units across Hawaii, DFW, and Tampa. Most bookings are insurance housing: families displaced by fire, flood, or a contractor who missed the date. Not passive. Not vacation-content. Just housing with operational drag.
Built to run Golden Glaze because the third-party stack was expensive and still missed the floor-level work. Daily checkins, scheduling, inventory, cash reconciliation, investor dashboards, and alerts — 7 modules across 7 locations. First version shipped in 3 weeks.
Built to run FlexStay Housing because the PM stack was built for vacation rentals, not insurance housing or multi-market portfolios. Guest messaging, pricing logic, turnover coordination, multi-market dashboards. Currently building inventory management and guest messaging modules.
Most "AI for SMB" stops at a chatbot. Mine doesn't. I run persistent AI agents that observe real operations 24/7, build per-store knowledge bases like new employees on shadow shifts, and graduate through a promotion ladder before they're allowed to talk to anyone outside a private review channel. It's an org chart, not a feature.
Each agent ships in observation phase. Daily digests into a private review channel. No public messages until I explicitly promote. Same way I'd promote a shift lead to GM — earned, not automatic.
Per-store profiles. Manager voice. Glossary of internal terms. Escalation patterns. Daily ingestion from Discord, email, POS, QuickBooks, and manager check-ins. The pipeline builds context the way shadow shifts do.
Who the agent can talk to is a list. What it can commit (no discounts, no menu changes, no hiring, no vendor terms) is a list. Same constraints you'd give a new employee on day one — written down, enforced in code.
Daily ops digests. Google review responses with voice guardrails. Catering order processing. B2B outreach drafts. Wholesale lead pipeline. Each task graduated explicitly when the agent earns it.
One decision, one number, one mistake from running 7 donut shops, 11 rentals, and the software stack behind both.
Every module in the stack that runs 7 donut shops — what each one replaced, what it cost to build, and how they connect.
Preview the artifact →How a $40 drawer variance gets caught same-day: register counts, deposit bags, variance flags, and the thresholds that page a human.
Preview the artifact →How 11 rentals across 3 markets answer guests without a front desk — drafts, approval gates, and the rules the system never breaks.
Preview the artifact →Every tool in the stack, what it replaced, and the exact build order — which leak to fix first and why.
Preview the artifact →The first-week build guide: pick one leak, describe it to AI, ship the first tool. The same loop the whole stack came from.
Preview the artifact →One tool a day for 60 days: the software that runs 7 donut shops and 11 rentals, torn down in order. Follow along on Instagram.
Follow the series →Hidden costs of self-employment, runway calculation, the real number you need before quitting.
Read the breakdown →Backwards from your exit date. Every milestone from T-12 to T-1, the non-negotiables for leaving with leverage.
Read the breakdown →How I planned my way to severance instead of quitting. The 5-step playbook nobody talks about.
Read the breakdown →When the layoff play isn't viable. The 4-rule script for resigning without burning bridges.
Read the breakdown →The first month after the W2 paycheck stops: calendar, cash, attention, and what not to touch yet.
Read the breakdown →What works, what doesn't, why I'd never recommend long-term rentals for cashflow. 7 vehicles ranked.
Read the breakdown →Built Not Bought — 60 days, one tool a day. Real numbers. Real wins. Real losses. The back-office parts most operators don't post.