This describes work that is in design. Everything below is what we are building, not something you can switch on today.
Tuesday morning. There is a listing to write for a unit whose details are already sitting in the system — square footage, floor, parking, the date it comes free — and the writing job is to turn those into three paragraphs that sound like your firm.
There are nine inquiries waiting, and each one needs a reply that is very nearly the reply you wrote for the last one, with two details changed.
And there is an application to read through, on a screen, where the four things you actually need are somewhere among the thirty that were collected.
None of this is difficult. Not one piece of it requires the judgement you were hired for. And all of it together is a morning, every week, spent by the person in the building whose time is worth the most.
The first thing we are building is the least impressive thing
The drafting layer lives entirely on your screen. It writes listing copy for a unit out of the details already recorded about it, in the voice your firm actually writes in. It proposes replies to the inquiries in front of you. You read what it wrote, you change what you want changed, and you press send.
Nothing leaves without a person pressing send. That is not a caveat at the bottom of this piece; it is the design. The drafting layer has no outbound of its own, no schedule, no autonomy about when a message goes. It writes, and then it waits for you.
What that converts is a morning of writing into a shorter session of approving. What it deliberately does not change is who is responsible for every word that leaves the building. That stays exactly where it is, with you, which is the reason you can use this on real correspondence next week rather than after a pilot.
The application on the screen is a third piece of that morning and it needs different handling from the other two — pulling out what somebody submitted is a more delicate job than writing about a unit, and it gets its own piece later in this series.
Why the boring thing goes first
This is chronologically the first capability on the roadmap, and it is not the most valuable one. The out-of-hours responder is worth more and it is being built second, on purpose.
Starting here is the only part of the programme where a firm can watch this software's judgement with a person standing between it and the world. You will see how it writes about your buildings, in your voice, on your actual mail, for as long as you want to — and every one of those drafts is a free look at what it would have said in a situation where nobody was checking. You form a view on evidence, not on a demonstration.
Two other things get made here, and everything later in the roadmap leans on them.
The first is a way of measuring whether any of this is any good. Not an impression — a record. Once there is a body of drafts and the sent versions beside them, there is something to test future changes against, and a new model or a new prompt stops being a matter of opinion.
The second is a picture of how your office actually writes, and it comes from the gap between what was drafted and what you sent. Your edits are the measurement. Nobody at your firm fills in a form describing your tone; you just correct drafts the way you were going to anyway, and the shape of those corrections is the most accurate description of your house style that exists.
That second one needs a careful line drawn through it, because it is easy to hear as something we have said we will not do. Your rules are written down and approved by you. Your voice is a matter of observation. We keep those apart deliberately: nothing learned from the way you edit a sentence ever becomes a rule about how your firm operates. A preference for short paragraphs and first names is style. Whether the deposit can be split over two months is policy, and policy only ever comes from the pack you wrote.
Anyone who tells you to start with the applicant-facing chatbot is optimising for the demo. Starting at the desk is the shorter route to the chatbot being good.
What it writes about, and what it never writes about
Two boundaries are worth stating, because they are what let this ship early.
Listing copy is about a property. It is assembled from what you recorded about that unit and written in your firm's voice, and there is no person in it at all.
A suggested reply is about logistics — what is available, when it can be seen, what the process is, what happens next. Also no judgement about a person.
Neither of those ever characterises an applicant, and that is not a restriction we ran into. It is the reason the drafting layer is safe to put in front of real correspondence first, while the parts of the roadmap that involve a stranger receiving something are built behind it.
The trade-off, named
Every draft still costs you the ten seconds it takes to read it.
Software that saves you a morning by adding an approval step to eleven emails is a straight trade, and it is worth saying so plainly rather than pretending the step is free. If the drafts are poor, that trade is a bad one — you will have added a reading task to a writing task and saved nothing.
Which is precisely why we would rather you judge it on your own eleven emails, on a Tuesday, with the send button under your thumb, than on anything we could show you.
What we are aiming at
Not an empty morning. A morning where the writing is already done and the work left is the part that needed you: reading nine drafts, changing three of them, and getting the rest of Tuesday back.
Releaser is available to property management firms in the United States and Canada today: one shareable link per property, your screening completed before anyone reaches your calendar, and self-serve booking for the people who meet the criteria you set.
The drafting layer is in design. This is what we are building it to be.