Peachy Adventure

Operations systems, built by hand.

Peach orchard in bloom, Woodbury, Georgia. March 2024, on film.

Open most days

I’m Jonathan Phillips, Head of Operations at Homegrown Storage, where we buy rural self-storage facilities and run them remotely; everything after a property is bought is mine. This stand is what I’ve built, how I think, and an invitation.

What’s fresh

  • Fall 2026. The stand opened. Four jars, two seconds, one farmer. More when there’s more.
  • Fall 2026. Tuning paid search ourselves for the first time, on a few test sites, measurement before spend; the plan behind it was benchmarked on the public operators’ filings.

Preserves

Finished work, one jar each. A claim and a number on the label.

  1. No. 1TransitionsSummer 2025 to Fall 2026 Closing to go-live: 28 days to next dayOpen the jar
  2. No. 2CollectionsFall 2025 to Summer 2026 Delinquency: 12.8% to 5.6%, while the sites nearly doubledOpen the jar
  3. No. 3Remote operationsSummer 2025 to Fall 2026 On site 20 to 40 hours a week, down to 1 to 3, run remotelyOpen the jar
  4. No. 4Tools my teammates useSummer 2026 A system of record you can ask in plain EnglishOpen the jar

Seconds

What failed, or what I would change. The part nobody else publishes.

Not all the juice is worth the squeeze.

The food-logistics startup

2022 to 2023

In my last year of college I talked our five-person capstone team into pivoting to a local-food logistics marketplace: small farms around Atlanta on one side, restaurants and wholesale buyers on the other, us in the middle with the truck. I built the delivery-routing model in Python and tied its costs into the financial case, built the two-sided storefront, ran the customer discovery, and after graduation ran deliveries with one truck and a part-time driver.

It failed. The margins were thin on both sides, a logistics layer on top of them was thinner, and every new city would have meant rebuilding both sides by hand. Scale had to be force-fed; it never compounded. What I would not build again: a thin-margin layer on thin-margin goods. What I took: let scale compound on itself or don’t build it, which is the test I now put every system through. And I have never learned more, faster, than I did failing at this.

The first migration sent the invoices early

July 2025

Six weeks in, mid-migration on my first five sites, the new property system started sending tenants invoices and payment notices before the sites were meant to be live on it. I found out from an airport, spent the layover on callbacks, and that night turned off every autopay and stored card in the seller’s old system by hand, so nothing could charge twice while two systems thought they owned the same tenants.

What I would change went straight into the four-phase checklist I wrote the next month, and it has been in every transition since: the seller’s automations go off and the new system’s messaging stays off until the cutover is done, as a named step with an owner. It was the most useful mistake I made that year, because the checklist that came out of it is the ancestor of the task engine in the Transitions jar.

The farmer

Who runs the stand.

I came at this sideways. Paid work on a 400-acre peach farm from fourteen: the retail shed and its cash, every machine with wheels, packing-house nights, 75 to 90 hours a week in season. Industrial engineering at Georgia Tech. A local-food logistics startup I co-founded in college and ran after graduation until it failed. First US implementation hire at an agricultural software startup, where I wrote the client dashboards in SQL, then hired and trained a developer and built myself out of that job, until the company wound down. A short pro bono stint building go-to-market for an agricultural fintech. Then a lunch with an old friend that turned into a pitch, and self storage, which I had never set foot in.

Every one of those ran the same way: learn the operation with my hands, build the system that runs it, hand it off, rebuild at the next threshold of scale. I scale with systems and AI instead of headcount. I learn the domain, set the judgment, and write the tools that take work off people.

  • Red clay at Gram’s. May 2023. Red clay at Gram’s. May 2023.
  • The truck, hood up, at home. January 2023. The truck, hood up, at home. January 2023.
  • Pruning vines, Fort Davis, Texas. 2023. Pruning vines, Fort Davis, Texas. 2023.
  • The road out to Fort Davis. 2023. The road out to Fort Davis. 2023.
  • A motorcycle on the bench, Woodbury. 2024. A motorcycle on the bench, Woodbury. 2024.
  • Joshua Tree, with the camera. 2026. Joshua Tree, with the camera. 2026.

Pull over

Five reasons people do. Pick yours; the subject line is already written.

I answer everything. Not always quickly.

pullover@peachyadventure.com

Same orchard, same week, at dusk.