BUILD & RUN — GENERATIVE AI
Staff already use AI in a browser tab, outside your systems.
Your team is already using AI in a browser tab and pasting the results back by hand. That is slower than it looks, produces something slightly different every time, and quietly moves your information through tools nobody has reviewed. This wires it into the places the work already happens.
Free · 20 minutes · no commitment
WHO THIS IS FOR
Teams where AI is already in daily use, unmanaged, with the results pasted back by hand.
THE PROBLEM
It also means none of it improves. Every prompt starts from nothing, nothing learns from what worked last week, and two people doing the same task produce two different documents. What looks like a productivity gain is largely a productivity shuffle — and the exposure is real whether or not anyone has thought about it.
WHAT IT LOOKS LIKE
Jobs
Job 2417 · change order 7
DRAFTED IN THE JOB FILE AT 3:12 PM · YOUR TEMPLATE, YOUR WORDING
Drafted from
2:40 pm
Site note from Luis: the client asked to move the kitchen wall 600 mm east. Two extra days.
SITE APP · 2 PHOTOS
2:55 pm
The extra framing and drywall, priced on the estimating sheet.
ESTIMATING SHEET
3:05 pm
The client’s email: “Go ahead, send me the paperwork.”
JOB INBOX
Change order 7 · draft
As agreed on site today, we will move the kitchen wall 600 mm east, reframe it and finish the drywall to match. Work on the change starts once you have signed this order.
- Price
- $3,860 plus HSTCheck
- Schedule
- Adds 2 working daysCheck
- Template
- Your change order, v3
- Prompt
- Shared change-order prompt, v4
Drafted in the job file, done
3:12 pm, from the site note, the sheet and the email
Dana checks price and wording, in progress
The two flagged values first
Dana sends it for signature, next
From the job file, under her name
Nothing reaches the client until a person has read it, and the job file is never used to train a vendor’s model.
What you get.
- Drafting and summarising built into the systems your team already opens.
- Output in your templates and your house style, not a generic voice.
- A review step wherever the output will reach a customer.
- Prompts held centrally, so quality does not depend on who happens to be asking.
How it works.
Priced against the scope, after a Diagnostic that is small on purpose. The shape of the work we can tell you now.
- THE SHAPE
- Built, then run. The prompt library is maintained as your templates and house style move.
- WHAT WE NEED FROM YOU
- Your templates, and a decision on which outputs need a human reviewer.
- WHAT YOU KEEP
- The prompt library and the templates, editable without us.
What this does not include.
Published because you should know before the call, not after the invoice.
- Publishing to a customer without a human check.
- Arrangements that let a vendor train on your customer data.
QUESTIONS
Is our data used to train someone’s model?
Not under the configurations we deploy. We use business-tier arrangements that exclude training on your content, and we put which ones in writing.
Will everything start sounding the same?
It will sound like your business, because it works from your templates and your past examples. For quotes and correspondence that consistency is the point, not a side effect.
MORE UNDER BUILD & RUN
WHERE THIS APPLIES
The industries whose page names Build & Run as one of the places we would start. Each one is written for that trade, with its own numbers.
Bring us one bad week.
Book a pilot call. Tell us where the week goes wrong, and we’ll tell you what an Operations Diagnostic would map and measure.
The first paid step has a clear scope and an end date, and the pilot after it is one workflow — not a platform, and not an open-ended retainer.
Free · 20 minutes · no commitment