Case study · built on our own floor

DMOS runs a real CNC shop. Ours.

Most manufacturing software is sold by people who've never set a tool offset. DMOS is different: it was built inside Dalloway Precision Engineering — a working West Midlands CNC shop doing milling, turning and surface treatments — and it runs the place every day. This is what changed, with the real numbers and how we measured them.

3 min 48 s
quotes used to wait 2–3 days
Median drawing upload → costed quote
~1 hour
from ~3 days of folder archaeology
ISO 9001 audit prep (owner-reported)
0 re-keys
was 4 disconnected tools
Quote → job → traveller → invoice

How we measured it: the quoting number is instrumented, not remembered — every quote run through DASQuote logs its pipeline with timestamps, and across the 44 timed quoting runs between 26 April and 25 June 2026 the median from uploading the PDF drawing to a costed quote was 3 minutes 48 seconds (nine in ten finished inside 14 minutes). The "2–3 days" before is how long a quote used to sit waiting for a quiet afternoon — owner-reported, not instrumented, and we label it as such. Audit prep is the owner's lived before-and-after, not a stopwatch study. "0 re-keys" is structural: one record flows from quote to invoice, and we'll happily show you that chain live. We used to publish an on-time-delivery percentage here too — we've taken it down until enough shipped jobs have flowed through DMOS for the sample to be honest. A number without a method is just decoration.

The shop before

Before DMOS, the shop ran on a tangle of spreadsheets, email threads and a quality system that was half in Word and half in a filing cabinet — the quote living in one person's head, the same job details re-typed into four separate tools between enquiry and invoice. It worked, just about, but every ISO audit meant days of folder archaeology, and none of that re-keying ever made a single part.

What we built — and why it's one system

The point was never six apps; it was one thread. A quote in DASQuote becomes an opportunity in DASCRM, becomes a job in DASBridge with its routing and materials, picks up inspection and NCRs in DASQMS, draws stock from DASFind, books labour through DASTime, and lands as an invoice — without anyone re-keying a thing. We felt every break in that thread because we were the ones living with it, so we closed them one by one until the shop genuinely talked to itself.

"I built DMOS because I was the one doing the quoting at nine o'clock at night. I'm not any more — and the audit evidence builds itself while the shop works."— Morgan Dalloway, owner, Dalloway Precision Engineering

The live trail — a small shop's real records

We're a small shop and this is early data — we'd rather show you a modest number that's real than an impressive one that isn't. Since the shop went onto DMOS in May 2026, every one of these is a live record we can open in front of you on a demo call:

What it actually changed

The quote number is the connected thread doing its job: the drawing, the part history and the pricing all in one place, so a quote that used to wait two or three days for a quiet afternoon now goes back the same day — and you win the work you used to lose to whoever replied first.

The audit number is capture-at-source. The NCRs, CAPAs, FAIs and calibration records are created as the work happens and stored as one linked chain, then export as a single clause-sequenced pack — so surveillance week stops being days of reconstructing evidence and becomes an hour of reviewing it.

And the zero re-keying is the dullest, most valuable change of all: the same job details used to be typed into four tools between enquiry and invoice. Now they're typed once, by the person who quotes it, and every document downstream — acknowledgement, traveller, delivery note, invoice — inherits them.

Why we publish the method, not just the number. We sell to precision engineers, who distrust a round number on sight — rightly. So every figure here says exactly what it is and how we got it: instrumented where we have the data, owner-reported where we don't, and removed where the sample is still too small. If a number looks too good, check the method underneath it. That's the point.
See it on your own jobs — book a 20-min demo