Bray Holcomb

Founder, High School Firewood · Freshman, University of Georgia

I run a firewood company. I built the software that runs it with AI.

I started High School Firewood at 16 with $500. It's now the #1 rated firewood company in the Southeast on Google. I built it around AI, with agents handling everything from a customer's first text to each driver's route.

Watch the 1-minute videohighschoolfirewood.comSee all 249 reviews on Google
Front desk booked Ms. Reyes for Saturday. Planner refused Truck C, capacity unknown, then planned 14 stops on 4 trucks with 2 reloads. Dispatch texted Mr. Cole his ETA. Books matched the cash from Truck B. Planning a day went from 3 hours to 10 minutes.
rated firewood company in the Southeast on Google
#1
Google reviews, all five stars
249
run rate this season
$200K
part-time crew members
29
to plan a delivery day
3 hrs → 10 min
on time across 105 deliveries
96%
Stop 01 · Try it

Plan a sample delivery day

Sample orders, not real customers. A simplified version of how my planner loads trucks: it fills each truck by what it can carry, schedules reloads, and won't approve a day when it doesn't know a truck's capacity.

8:00 AMSample day
Delivered
0/0
Stacked
0
On the road
0

      Customer texts

      0 sent

        Trucks

        Capacity in pieces of wood

        Clear a capacity to see the planner refuse the day.

        The day

        Orders

        #ProductPcsTruckLeaves barnArrivesStatusRemove

        One order always rides on one truck.

        Stop 02 · The barn

        One minute at the barn

        Four members of the High School Firewood crew in the barn
        Stop 03 · Proof of work

        The system

        I built an operator app and a driver app with AI coding tools. They've handled real deliveries since August 31, 2026, taking orders from the first customer text through delivery and payment.

        Two crew members going over a delivery on the driver app
        Stop 04 · The route here

        How I built it

        1. Sep 2023

          Started the company at 16 with $500, delivering wood from my grandpa's land for gas money.

        2. 2024-25

          Used ChatGPT to design a four-tier pricing ladder. Orders went from 71 to 229 in one season.

        3. The ice storm

          An ice storm sent orders in faster than I could route them by hand. Planning one delivery day took about 3 hours.

        4. Summer 2026

          Built the driver and operator apps with AI coding tools. My first launch broke because the planner ignored how much a truck could carry, so I rebuilt it around loads.

        5. Aug 31, 2026

          Went live on real deliveries. Day planning dropped to about 10 minutes.

        6. Fall 2026

          Added AI that answers customer texts, plans every route, and confirms each delivery, so the business keeps moving while I'm at UGA.

        The High School Firewood barn with cut logs stacked in the yard
        A pickup and trailer hauling logs into the wood yard
        A full rack of firewood stacked neatly at a home
        Stop 05 · Other work

        Other work

        This summer I worked two finance internships. I also built a tool for UGA's student-managed fund, and I test business ideas with my co-founder. Names and numbers are changed to keep client work confidential.

        • Joseph Breland Capital · M&A Analyst Intern · Summer 2026

          What buy-side advisors actually charge

          I called 50+ advisory firms, weighted 121 sources by quality, and built five fee schedules for $20M to $50M buy-side deals. The firm now uses them.

        • Joseph Breland Capital · M&A Analyst Intern · Summer 2026

          41 new targets for a 574-company pipeline

          I screened 200+ companies against a buyer's mandate, added 41 new targets to the firm's 574-company pipeline, and tested whether the lead target could actually close.

        • Animated illustration: 561 client households fly into gift tiers, each gift is packed into a box, and the shipped boxes fly to a stack while a counter climbs to 146.

          Harbor View Private Wealth · Wealth Management Intern · Summer 2026

          Thank-you gifts for 561 households

          I built the tracker the firm used to send thank-you gifts to 561 client households, from writing the card to packing, shipping, and tracking the box.

        • Harbor View Private Wealth · Wealth Management Intern · Summer 2026

          When a referral had no policy

          A referral came in and the firm had no policy for paying it. I researched peer firms and designed the referral compensation program. The firm adopted it.

        • UGA Student Managed Investment Fund · Since October 2026

          Connecting Claude to Bloomberg

          I connected Claude to the Bloomberg Terminal for UGA's student-managed fund, which admits only juniors and seniors. Members use it and my setup guide to build stock pitches.

        • Three business ideas each get a rule that can kill them; one is killed, routing is tested and computed routes drive 14% fewer miles, but the scorer check reaches 86% against a 90% pass mark set in advance, so it is not passed and the bar is kept.

          LATS · Research lab with my co-founder · Fall 2026

          We don't move the pass mark

          My co-founder and I test business ideas before we build them. Our first test cut my firewood company's delivery miles 14% but failed its accuracy check, so it runs in the background for four weeks.

        Stop 06 · Next route

        What I want to build next

        I want to train an AI agent on our real delivery data. It'll practice thousands of Saturdays in a simulator, learning to make more profit per driver hour. If it works, I want to bring it to other seasonal service businesses.

        Why the Academy

        I've spent the last few years figuring things out as I go. I've made AI work for one business, but I have a lot to learn if I want to make it work for thousands.