Side project Mobile tire service In production

Breaux's Roadside & Tire

A business run on software I created, end to end

A mobile tire business had been running for almost a year on less than a job a week, with no website and no way to tell which calls were worth paying for. I built the site first, then the tracking behind it. As calls and revenue grew, the paperwork became the bottleneck: six manual systems that all had to be reconciled by hand. Nothing off the shelf fit what the owner needed, so I wrote the web app the business runs on every day. Once the calls were measured, the next question was how much work is out there at all, so I tried to size the market.

The app's Jobs tab: a searchable, sortable list of job cards, each showing the request, call time, tire category, and status.
The jobs list. Everything else in the app hangs off the job record.
The app's Look up tab: a running stock count above a searchable list of tire sizes, each row showing the quantity currently in stock.
Inventory in the same app, so stock is checked against the job rather than a separate spreadsheet.
A map of Mississippi with every federal-aid highway drawn as a colored line, dark red for corridors carrying 9,000 or more tractor-trailers per day down through orange and pale yellow for under 150. I-20, I-55, and I-10 read as the heaviest lines on the map.
Truck corridor exposure: how much work is out there that hasn't called yet.
The customer-facing breauxs-roadside.com site, the landing page ads point at.
breauxs-roadside.com is where the ads point and where call tracking starts.
10× Ad spend increase Based on increased calls and revenue, now traceable to source
6→1 Manual systems replaced by software I wrote Call-source tracking, quoting, invoicing, inventory, expenses, and accounts, all on one job record
Daily Production use of that app Built and operated solo, against a live database
Chronological Steps

From no calls to custom software

This wasn't a bunch of separate jobs. It was one problem worked in order, and each step showed me more clearly what the business needed built.

01 No website, no tracking, almost no calls Ads pointed at a Google Business Profile. Nowhere to place conversion tracking.
02 Built customer-facing site, rebuilt the advertising breauxs-roadside.com, then Google Ads search and Performance Max, plus Local Services Ads.
03 Made calls measurable end to end Joined three platforms to the one thing none of them knew: did the call become a job?
04 Once calls were measured, the owner spent more, and the business grew From reluctance to double the budget, to over 10× the original spend in one to two months.
05 Created the software the business now runs on No off-the-shelf tool fit the workflow the owner wanted, so I built one: a Cloudflare Workers and D1 web app putting call-source tracking, quoting, invoicing, inventory, expenses, and fleet accounts on a single job record. In daily production use.
06 Asked how much work is out there, not just where calls came from Planning past the calls he already had meant estimating a number nobody publishes. A Python pipeline over federal highway and carrier data ranks Mississippi corridors by truck exposure and maps the fleets within 75 miles.
Case Studies
Components

Shipping and running the app

Writing the app was only half the work. The other half is keeping a live system safe to change.

  • Cloudflare. Static customer-facing site, Workers, D1, DNS, and redirect rules.
  • Data layer. D1 schema design, migrations, and backups behind quoting, invoicing, inventory, expenses, and fleet accounts.
  • Safe deploys. An integration suite against real D1 that must pass before merge, migrations written to be additive, and a hook that blocks writes to the production database without a fresh backup.
  • GitHub. Organization setup, repositories, and deploy pipelines, with owner-level access held by the business rather than by me.
  • Continuity. Accounts, admin access, and email are all owned by the business, start to finish.
  • In progress. Call-capture automation with OpenPhone and structured extraction from call transcripts.