DISPATCH DESK · PROPERTY MAINTENANCE & RENOVATION

Dispatch

The desk that never goes to lunch.

Your crews are in the field. Your projects are in a spreadsheet. Your tenants are texting your personal phone. Dispatch puts all three on one screen, picks who to send, and writes the messages before you ask.

Fully interactive · fictional data · walkthrough first

Dispatch — the desk
3Unassigned
1On the way
3On site
2Complete
8am9101112pm12345
Marcus BellPlumbing
Faucet swap
Tasha RuizElectrical
Panel + GFCICeiling fan
Devon CarterHVAC
Tune-upThermostat
Luis OrtegaGeneral
Cabinet doors
Priya NairPaint
Unit turn, 2A
WITHOUT IT
  • Three apps and a group chat, none of which agree
  • “Where’s my guy?” calls you answer by calling your guy
  • The same update typed twice, slightly differently
  • A lockbox code sitting in a tenant’s text history
  • Your personal mobile number, in forty tenants’ phones
WITH IT
  • One board, one thread per job, one business number
  • Tenants told before they think to ask
  • One note, split correctly, sent to both sides
  • Codes and pricing never reach the tenant’s copy
  • Your number stays yours
THE THING TO TRY FIRST

One note in. Two texts out.

You write what you’d write anyway: “Arrival is moving to 2:30. Parts approved up to $250. Lockbox code is 4471.”

Your contractor gets all of it. Your tenant gets the part that concerns them — no code, no pricing, no internal detail. You see both drafts before either one sends.

Marcus gets

Marcus, update on 412 Glenwood Ave SE: Arrival is moving to 2:30 PM. Parts are approved up to $250. Lockbox code is 4471.

Dana gets

Hi Dana, an update from Oakhurst Property Works about your water heater leak: Arrival is moving to 2:30 PM. Reply here with any questions.

WHY IT PICKS WHO IT PICKS

Two AIs, two jobs.

Run reads a raw request and answers in fixed shapes rather than sentences: which trade, urgent or not, a fit score for every crew — each with a confidence and the milliseconds it took.

Dispatch turns those numbers into words. It names a crew, tells you why in one line, and drafts the texts. You approve. Nothing goes out on its own.

Marcus BellBest fit
SCORE 0.8789% conf214ms
  • Plumbing is the right trade
  • Working Eastside today, short drive
  • Open 1:00 PM to 3:00 PM
CHOICE Plumbing91%YES/NO urgent96%
HOW RUN DECIDES

It never writes a sentence. That’s the whole trick.

Asking a model to describe a job in words gets you words you then have to read and judge. Run answers in one of three fixed shapes instead, so the answer can be checked at a glance rather than interpreted.

CHOICE Plumbing

A choice from your list

You hand it the five trades you actually have. It has to hand one of them back. It cannot invent “Roofing” for a company that doesn’t do roofs, which is exactly what open-ended text does when it isn’t sure.

YES/NO urgent

A yes or a no

Urgent, or routine. Nothing in between, nothing hedged, nothing that needs a second read to work out which way it landed.

SCORE 0.87

A number out of one

How well one crew fits one job. Built only from things you could have checked yourself, which is what makes it arguable rather than magic.

And the score isn’t a black box

Every crew starts at 0.50. Four facts move them:

Right trade for the job+0.50
Wrong trade−0.15
Already working that project today+0.20
Already in that part of town+0.14
Actually free in the window+0.18
Booked during the window−0.55

Being booked is the heaviest penalty on the board, because a perfect match who can’t go isn’t a match. And every answer carries a confidence, so a 41% guess looks different on screen from a 94% one — you can see when it’s reaching.

HOW IT WORKS

Three steps, and you only do the last one.

01

A request lands

A tenant texts, a client emails, or a project task comes due. It goes in the queue, classified by trade and marked urgent or routine.

02

Dispatch names a crew

Every crew gets scored on trade, on whether they're already at that site, on drive time and on whether they're actually free. You get a ranked pick and one line on why.

03

You approve, it goes out

Two texts are already drafted — the full detail for your contractor, the plain version for your tenant. Edit anything, send both, and the thread starts.

WHAT’S INSIDE

The field side and the project side, finally on one screen.

A board, not a group chat

Every crew, every hour, side by side. Who is free at two o'clock stops being a question you answer from memory.

Status without phoning anyone

The contractor taps a link in their text. On the way, on site, complete. The board moves and the tenant gets told.

Arrival texts that send themselves

No app for the crew to install, no portal for the tenant to log into. It's a text message, because that is what people answer.

Projects and jobs in one place

Estimates, budget against spend, phases from lead to punch list — and any task on a project can be pushed straight to the board.

Intake that sorts itself

Paste a raw tenant message. It comes back classified by trade, flagged urgent or routine, and sitting in the queue.

One number, two sides

Contractors and tenants never see each other's phone numbers, and every reply lands back in the same thread.

QUESTIONS

The ones that actually get asked.

Does my crew need an app?

No. Everything a contractor does happens in a text message — accepting the job, marking themselves on the way, marking the job complete. There is nothing to install and nothing to log into.

What stops a tenant seeing something they shouldn't?

You write one note. The version sent to the tenant is generated separately, with access codes, pricing, parts approvals and internal notes stripped out. You see both drafts before either sends, and you can edit them.

Do texts go out on their own?

Arrival and completion notices do, because they are triggered by the contractor's own tap and have nothing in them to get wrong. Anything you write is drafted and held until you send it.

Where do replies go?

Into the job's thread, next to everything else about that job. Both sides message one business number, so nobody ends up with your mobile and no conversation lives in someone's personal phone.

Is the AI making the decisions?

It ranks and it drafts. It does not assign anyone or send anything you have not approved. Run's scoring is visible — you can see what it weighed and how confident it was before you accept the pick.

Is this a product I can buy?

Not yet. It is a working concept built by martin.builds to show what this looks like when it's real. If your operation works this way, the build is the conversation.

Dispatch

Open it and dispatch a job.

Three jobs are waiting for a crew. Assign one and watch the texts write themselves, or paste a raw tenant message into the queue and see it classified.

See the demo

Opens after a short walkthrough · book below