When an engineering automation is worth building
Engineering automation, in this studio, means a purpose-built workflow around drawings, calculations, device data and technical documents — not a generic business app. The test is simple: is there a stable rule, a stable schema, and a cost of error high enough that a human should not retype the same fields? If yes, specialized software and engineering automation is in scope. If the “rule” is actually a senior person’s judgement on every sheet, automation will fossilize a guess.
How the investigation is actually run
Sit with the person who does the task and count stable steps versus judgement steps. Stable: the same title-block fields, the same family of devices, the same mapping between two schemas. Judgement: whether this feeder is special, whether this note overrides the typical. Automate the first class. Leave the second on the sheet. If the client will change the standard next month, you do not have a rule, you have a wish.
Name the source of truth and the failure mode. Silent wrong schedules are worse than a tool that refuses to run. A utility that demands a second career in wizards has already failed the operator test. Prefer a small program beside CAD over a platform conversation that nobody can maintain. A weekly spreadsheet is honest when the mapping is still moving.
The scoping memo is a yes, a no, or a narrower slice — with the objects named. It does not promise recovered hours. Studio Work examples stay examples until someone engages a build.
When it is worth it
Repeated mapping between two applications (see connecting engineering applications and the ETAP–AutoCAD data bridge). Tagging, layer discipline, schedule extraction. Routing of submittals by document type with a human still on the exception path — technical document automation, not a slogan. Internal calculators that encode a house method so two people do not diverge.
When it is not
A one-off custom drawing. A client who will change the title block standard next month. A process whose only specification is “make it like last time.” Tools that require the engineer to become a full-time operator of a brittle script.
How we decide
Count the stable steps. Name the source of truth. Name the failure mode if the tool is wrong (silent bad schedule versus an obvious crash). Prefer a small utility beside the drawing — internal engineering utilities — over a platform pitch. The case for a larger system of record is separate: EE-Platform.
The conclusion of a scoping conversation is a yes, a no, or a narrower slice. It is not a promise of hours saved. Related work on the studio site lives under Work unless engaged as a scoped build.