AI by industry

AI for manufacturing and construction

The quote takes three days and goes to whoever sent theirs first. Work orders go out by phone, the site report lands in a group chat and disappears. That is work that belongs back in a system.

In manufacturing and construction AI is worth most before the job starts — on the take-off and the quote. From a specification, a drawing or a list of items comes a costing with your norms and your prices in hours instead of days, so you can still answer while the job is open. The second part is the site: a work order on a phone, a report with a photo and materials used, and records that show planned against actual while you can still do something about it.

What actually gets automated

Take-off and quote

ProblemThe specification arrives as a PDF or on paper, and the costing gets typed into Excel over three days.

SolutionItems are read out of the document, matched to your norms and prices, and produce a costing with margin that you check.

DeliveredCosting in a spreadsheet and a PDF quote on your letterhead.

Work orders on a phone

ProblemThe order goes out by phone, so nobody knows who did what or how long it took.

SolutionThe crew gets the order with items and materials, and closes it with hours used and a photo.

DeliveredField view in the browser, no install.

Site reports

ProblemPhotos and notes end up in a group chat and nobody can find them a month later.

SolutionThe daily report is filled in from a phone, with photos and timestamps, and files itself against the job.

DeliveredSite diary with an archive per project.

Materials and waste

ProblemOne thing was ordered, another used, and the difference shows only at the end of the job.

SolutionUsage is recorded against the work order, so variance from the norm shows during the job, not after it.

DeliveredMaterial records per project with planned against actual.

Certificates and deadlines

ProblemTest certificates, warranties, equipment inspections, working-at-height training — a miss means a stoppage or a fine.

SolutionEvery deadline in one place with a warning and a named owner.

DeliveredDocument register with reminders.

Catalogue and technical documentation

ProblemA customer asks for a datasheet, and it exists in three versions on three computers.

SolutionOne documentation source with search by meaning — you ask a question and get the right document and page.

DeliveredSearchable document base with version control.

Why the quote, and not the production line

People expect AI in manufacturing to mean robots and failure prediction. For a company of twenty that is not the first step, it is the fifth. The first is the quote — because that is where you lose work that was already on the table, with no investment in equipment.

What stays with the engineer

  • Checking every costing before it goes out — the machine reads the document, it does not carry the liability.
  • The technical solution and any departure from the design.
  • Risk and site safety assessment.
  • The relationship with the client and negotiating deadlines.

How AI gets introduced into a company

  1. 01

    One hour of conversation

    We measure how long the path from enquiry to a sent quote actually takes.

  2. 02

    One job, not all of them

    Almost always the take-off and the quote — that is where the loss is.

  3. 03

    Two to four weeks to launch

    Built on your norms and your supplier prices.

  4. 04

    Two weeks running in parallel

    Costings are produced both ways until they agree on real jobs.

  5. 05

    The next job

    Then work orders and site reporting.

FAQ — Manufacturing and construction

Does it read drawings?

It reads specifications, item schedules and written descriptions reliably. Drawings partially, and always with a check — on a drawing an error is not forgiven, so nothing relies on automation there.

Our norms are in someone's head, not in a table.

Then the first step is writing them down. That is a few days of work and it is worth doing with no software at all — a company whose norms live in one head cannot grow.

Our site crews do not have good phones.

The field view runs on older handsets and over a weak signal, buffering when the connection drops. If it demands newer devices, it is built wrong.

We have an ERP. Does this go through it?

It goes alongside it. The ERP stays the source of truth for stock and finance, and this layer solves what ERPs traditionally do badly — fast quoting and capture from the field.