Two inks
Stock 120gsm warm
±0.13mm · in register

Short runs of useful software.

Fig. 1 — Title plate. Two inks,
hand registered, one sheet.

Drag registration in the docket
above to pull the plates apart.

I am one person with a small press. I build automations, document pipelines and websites — made to order, in short runs, by hand. No platform, no subscription, no team to feed.

What
Automation · Document processing · Websites
Who
One person, brief to handover.
Nobody to escalate to.
Run length
Days and weeks. Not quarters.
Handover
Source, notes, a walkthrough. No lock‑in.

Fig. 2 — On the
economics of a
short run.

Most software is printed in editions of millions.
Yours doesn’t have to be.

A script that renames forty thousand files correctly. A pipeline that turns a month of invoices into one clean spreadsheet. A page that loads before you have finished blinking. These are small jobs.

Small jobs are badly served by big software. They get a subscription with two hundred features you will never open, or a six‑week engagement with three people on the call and someone taking notes about the notes.

What a small job actually needs is one person who will set the plates, pull a proof, run it once while you watch, and hand you the finished thing.

So that is what this is. The talking, the building, the testing and the handover, all the same pair of hands. There is nobody to escalate to, which turns out to be the feature.

Fig. 3 — Proofs pulled
before this press.

Public work
Experience with automation
Before Small Runs, I built automation for a regulated pharmaceutical production lines.
That work is not public, but the kind of problem — exacting, high‑stakes,
no room for a script that fails quietly — is one I have already spent years solving.
Job ticket — unassigned No. ______ · Received ____ / ____

The ticket is
blank on purpose.

Those three plates are the work I set most often. They are not the whole catalogue.

If you can describe the fiddly, repetitive or quietly expensive thing you keep doing by hand, there is usually a way to hand it to a machine. Tell me what it is and I will tell you honestly whether it is worth building — including when the answer is no, keep doing that yourself.

Describe a custom job

The thing you keep doing by hand
How often it happens
What it costs you when it goes wrong
What you have already tried
Deadline, if there is a real one
One sheet, one job · nothing is set until we have talked

How a job runs

Four stations · feed to delivery

Station 01

Brief

You describe the job in plain words. I tell you what it will take, what it will not fix, and whether it is worth doing at all.


Station 02

Proof

A small working version, early and unpolished, running on your real data. Not a picture of the thing — the thing.


Station 03

Run

Built properly and handled failure by failure, then run end to end once while you watch it happen.


Station 04

Handoff

Source, plain‑English notes and a walkthrough. It runs on your machines. Nothing phones home, nothing expires.