Case Study
How a radio rental company outgrew its no-code database — and what we built instead.
DMS rents and sells two-way radios. Its customers are blue-collar businesses — trucking, towing, construction — and large annual events like the NYC Pride parade, where hundreds of radios go out the door for a weekend and have to come back.
The business runs on knowing exactly where every radio is. Who has it. When it's due back. Whether it's reserved for next weekend's event. With more than 25,000 items in the fleet, that is not a spreadsheet problem.
For years DMS ran on Knack, a no-code online database. It worked when the business was smaller. By 2026 it was holding the company back in ways that cost real time and real equipment.
Knack keeps at most one month of change history on its mid-tier plan, and none at all on its starter plan. When a radio went missing, there was no way to see who last touched it or which order it went out on. Today about 1,270 items — five percent of the fleet — are marked missing or stolen. A permanent trail would not have found all of them, but it would have found some.
Knack treats a barcode scanner like a keyboard typing into one box. For every unit you click the box, scan, click the next box, scan. For an event order with 200 radios going out Friday and coming back Monday, that is hundreds of extra clicks and a lot of chances to miss one.
A mistaken batch edit or an accidental delete was permanent. Restoring a backup is a paid service.
Nothing stopped the same radio from being assigned to two overlapping events. Double-bookings had to be caught by eye.
The fleet alone is over Knack's 20,000-record starter limit. With the links between items and orders, it is over the 50,000-record mid-tier limit too. Staying meant moving to a $300-a-month plan just to hold the data — and still only getting two months of history.
Knack is a fine tool for a simple database. DMS had simply outgrown it. Once a business needs scanning workflows, permanent history, and reservation rules, it is paying a subscription and writing custom code on top anyway.
DMSCommand is a web application built specifically around how DMS actually works. It runs in any browser, on the shop computer or a phone in the warehouse.
Rentals, sales, event orders, repairs, activations, transfers, and demos share one numbering system and one workflow. Big events can be split into sub-orders. Today it holds 625 orders going back to January 2020, including 161 event orders.
Staff open an order line, pick up the scanner, and go. Each scan adds the item and checks it on the spot: Is it the right model? Is it actually available? Has the planned quantity been hit? Wrong items are flagged in red before they leave the building.
The same scanning flow creates new inventory when a shipment arrives — scan a stack of new radios and they are all in the system.
Every change to a radio — assigned, returned, sold, repaired, marked missing — is recorded with who did it, when, and why. There are 112,000 entries so far, including a trail rebuilt from the old records back to 2017, and they never expire. Click any item and read its whole life.
Assigning an item to an order reserves it. Processing a return releases it. Each line shows planned versus scanned at a glance, and if two orders collide, the app flags the conflict and offers a swap tool to fix it.
Deleted records go to a recycle bin for 90 days. Recent batch changes can be rolled back from a notice in the corner of the screen. Nothing is gone for good because someone clicked the wrong button.
Purchase orders track what was ordered versus what arrived. Invoices and returns link back to the original order. Service calls, tasks, and contacts live in the same place. Every table lets each person choose their own columns, filter, edit in place, and batch-edit — from the normal screen, not an admin panel.
Nothing could be lost in the switch. We carried over the entire business from Knack in one pass: more than 24,000 inventory items, 350+ contacts, 275+ models, and every order, return, and purchase order going back to 2020. Orders, returns, and purchase orders keep their original Knack ID, so anything referenced in an old email or paper file can still be found. The item history was rebuilt from the old records too, so the trail on a radio starts in 2017, not on launch day. The import can be re-run safely without creating duplicates, which meant the team could keep working in Knack right up until cut-over day.
An operations system is only useful if it is trusted. Before every change goes live, hundreds of automatic checks run against the app — including full click-through tests of the real workflows. After every update, the live site is verified automatically. Three scheduled jobs run every day and report in on an admin page, so a silent failure can't go unnoticed.
Figures from the live system, October 2026. Screenshots use demonstration data, not customer records.
Nine weeks from first checkpoint to daily use, August to October 2026. 587 saved checkpoints. 81 safe changes to the data structure as the system grew. Built using AI development tools, directed by someone who ran DMS's inventory, shipping, and database for eight years — so every screen was designed by the person who would be standing at the scanner.
Tell us what's breaking. We'll tell you in plain English whether a custom system makes sense — and what it would cost.
Start a Conversation