AI that works the way you do
Hand off the work that wears you down.
A friendly, no-pressure guide to setting AI up around your real day, one task at a time, so you get hours and energy back for the work you enjoy.
Tap the model · swipe
What changes
A lighter day, in your own voice
A draft waiting, not a blank pageReplies, reports and summaries start half-done.
Written the way you'd say itIt learns your tone, your rules and the people you work with.
Time back for what mattersCalls, hands-on work, or simply an earlier finish.
You stay in chargeNothing goes out until you've given it the OK.
How it works
Four steps, one task at a time
Tap a step for the details.
Built around your work, not a template
Most AI setups start with a tool and hope it fits. This one starts with a task you already do, so what you get fits your day from the first week.
It's the same reason a good contractor looks at the job before quoting it: a quick look at the real thing makes everything after it simpler.
What that means for you
Nothing gets set up that you'd have to undo later, and every hour you put in builds on the last.
What you keep
What you walk away with
A map of the taskHow it works today: what comes in, the calls you make, what goes out.
A working setupAn assistant set up for that one task, using your own material.
Your preferences, written downYour voice, rules and limits, yours to keep and reuse anywhere.
An honest readWhat it saved, what it didn't, and whether it's worth going further.
What I'll ask of you
Your part is small
Five minutes with this guideYou're already most of the way through it.
Ten to fifteen minutes on the intakeOne question at a time, saved as you go.
One short walkthroughYou show me a task, live or on a quick recording.
A few real examplesRecent work, including one you'd redo.
How it learns you
It learns you the way a new hire does.
A little patience up front, then it remembers how you like things done and keeps getting better.
Tap the curve · swipe
A lot like training a new hire
On day one a new hire doesn't know your voice, your owners, your guests, or how you like things written up. You spend the first few weeks correcting them, and honestly it can feel slower than doing it yourself.
Then the corrections stick, and the work starts coming back right the first time. An AI assistant follows the same curve, except it never forgets a correction once it's saved.
What one correction looks like
A guest asks
"Any chance we can check in early?"
First draft
"Of course! Early check-in is no problem, see you at noon."
You correct it once
Never promise early check-in. Offer to text them the moment the unit is ready.
Every draft after that
"We can't promise early check-in, but we'll text you the minute the place is ready so you're not waiting around."
That correction is now a saved rule for every guest who asks.
The learning curve
Slower first, then it pays you back
An illustrative shape, not a measurement. How soon you reach break-even depends on the task and how often it comes up.
What gets saved
Every correction becomes something it keeps
Your voice
How you talk to guests, owners and vendors.
Your rules
The calls you make without thinking.
Good examples
Past work you were happy with.
Your limits
What it should never say, promise or touch.
These are plain written documents that belong to you, and they come with you if you ever switch tools.
A short dip, then a long payoff
The first couple of weeks can feel like more work, not less, and that's completely normal.
Knowing it's coming is what carries people through to the good part, where each week takes a little less out of you than the one before.
Time & energy
Saved energy counts, even when the clock doesn't move.
Approving a draft takes far less out of you than writing one from scratch.
Tap the model · swipe
Writing versus approving
Your job changes from writing to approving
Some tasks only take ten minutes but drain you. Handing over the first draft means you start from something instead of a blank page.
Writing it yourself
Find the contextBlank pageDraftFix the toneSend
Approving a draft
Read itAdjust a lineSend
Where the freed-up attention goes
While the assistant drafts one thing, you can be doing another, and the attention you free up goes somewhere better.
- The call you've been putting off
- Hands-on work only you can do
- Ending the day with something left in the tank
My promise to you
If checking a draft ever takes as long as doing the task yourself, I'll say so and we'll pick a better one. You only keep what genuinely helps.
Your numbers
What one task costs you in a year
Minutes each time
15
Times a week
10
Minutes to check a draft
5
hours a year back, about working days.
An estimate from your own numbers. It leaves out the teaching time in the first few weeks.
Is it a good first task?
Picture one task as you answer.Happens weekly or more
Follows a pattern you could teach
You have an example done well
Checking a draft beats writing one
Mistakes would be easy to catch
Where to start
Good first tasks, and ones to save
- Replying to common guest questions
- Turning turnover or inspection notes into a clean report
- Summarizing a week of messages into a to-do list
- Drafting responses to reviews
- Tidying invoices, receipts or expense notes
- Anything where a mistake costs money and is hard to spot, like pricing
- Refunds, disputes or anything legal
- Work that's different every single time
- Tasks only you understand and haven't explained to anyone yet
Questions people ask
Straight answers to fair doubts.
The questions people usually ask first, answered plainly. Good questions mean you're taking this seriously.
Tap the model · swipe
Before we start · tap to read
Along the way · tap to read
Consultant notes
Walk it before you quote it.
How to steer the first conversation when a client doubts what's possible.
Tap the model · swipe
Why clients push back
Most pushback comes from being asked to trust something people can't picture yet. Some have been oversold by AI hype, and some tried a tool once with no context and got generic results.
Arguing with either doesn't work, but one real example of their own work in front of you does.
The rules
Four things to hold to
No suggestions without inputs and outputsA guess spends credibility you'll need later.
Anchor every question to a real instance"What do you want AI to do?" only gets wishes.
Show one win on their own materialIt persuades more than any explanation.
Watch for twenty minutesIt beats two hours of talking. Shadow or get a recording.
Line that lands · Setting the frame
I can't give you a repair quote over the phone without walking the unit, and this works the same way. Let me watch you do one of these, or send me the last three, and I'll come back with something specific.
Line that lands · The training time
Think of it like a new hire. The first few weeks you're correcting it, then the corrections stick and the work starts coming back right the first time.
Line that lands · The energy win
It's the difference between writing the email and approving the email.
Line that lands · Earning trust
If checking its work takes as long as doing it yourself, that task isn't ready, and I'll tell you so.
Sending this to a client
Share it firstThe page is private until you share it from its Share menu.
The plain link opens the client guideThese notes stay tucked behind the menu.
Add #intake to the end of the linkIt opens straight to the intake form.
Interview
Questions that get you real material.
Five groups of questions worked roughly in order, plus the signals to watch for while they work.
Tap the model · swipe
Jump to a group
Group 1 of 5
Find the pain
What did you do this week that you'd hate to do again?Surfaces repeat pain faster than asking about goals.
Which tasks drain you even when they're quick?Finds the energy wins people never list as time wins.
What do you keep putting off?Avoided tasks are often the ones a first draft would unstick.
Group 2 of 5
Walk the work
Walk me through the last time you did it, step by step.Anchors to a real instance, not the idealized version.
Where did the information come from?Your inputs, which decide what it needs access to.
Who got the finished version, and what did they do with it?Defines the output and who it has to satisfy.
What did you decide along the way?The judgment calls are what turn into rules.
Group 3 of 5
Get examples
Can you send me the last three of these?Real inputs and outputs, side by side.
Show me one you were happy with and one you had to redo.Good and bad examples teach faster than any description.
What do you tell a new person when they start doing this?Existing training notes are ready-made rules.
Group 4 of 5
Set the limits
What would be a disaster if it went out wrong?Tells you where a person must approve every time.
What should it never see?Sets the data scope before anything gets connected.
Group 5 of 5
Define success
If this worked, what would you do with the freed-up time?Their own words for the value, which you'll use later.
How would you know it's working?A measure they agree to now is one they can't move later.
While you watch
Signals worth writing down
Each of these usually points to something an assistant can take over.
- Copying the same details from one app into another
- Retyping information that already exists somewhere
- Checking three places to answer one question
- Rewriting a message just to get the tone right
- A sigh, a stall, or "I hate this part," which is the energy cost showing
Pushback
What they say, and what's underneath it.
Six objections you'll hear, the worry behind each, and what to try.
Tap the model · swipe
1 of 6 · They say
“Just tell me what AI can do for us.”
UnderneathThey want reassurance it's worth their time before they invest any.
TryName three general strengths (drafting, summarizing, sorting), then ask which one sounds like something they did this week. Now you're talking about a real task.
2 of 6 · They say
“Give me the full plan and a price up front.”
UnderneathFear of open-ended cost, and of being upsold later.
TryOffer the fixed-price first step with a clear finish line. The full plan comes after the walkthrough, when it can actually be accurate.
3 of 6 · They say
“I tried it. It's garbage.”
UnderneathThey used it cold, with no examples, rules or context.
TryRerun their last bad attempt in front of them with one good example and two rules added. Seeing the difference does the arguing for you.
4 of 6 · They say
“We don't have time to teach it.”
UnderneathThey're picturing endless setup across their whole business.
TryShrink it to one task and a few short correction rounds, and show them the saved rules growing each time.
5 of 6 · They say
“Is this safe with our guests' information?”
UnderneathA legitimate concern, so treat it as one.
TryDecide what's in and out of scope before anything else, put it in writing, and check together how the chosen tool handles data.
6 of 6 · They say
“This is going to replace my people.”
UnderneathLoyalty to staff, or worry about how the team will react.
TryFrame it around first drafts and sorting, and bring the staff member who does the task into the pilot. They'll catch problems you'd miss.
First step
Price the walkthrough as its own step.
A small fixed-price pilot with a clear finish line, then a go or no-go.
Tap the model · swipe
Why discovery shouldn't be free
Discovery is the hardest and most valuable part of the work. Giving it away to win the project teaches clients to expect the thinking for free.
A small fixed-price first step also answers their fear of open-ended cost, because they can see exactly what they get and where it ends.
A suggested shape
- WEEK 1WalkIntake, one walkthrough, samples collected. You write the task map.
- WEEK 2BuildSet it up with their rules and a few good examples, then run the first correction round together.
- WEEKS 3–4Live with reviewReal work, approved before it goes out. Every correction saved.
- CLOSEGo or no-goA short note on what it saved, what it didn't, and whether to expand, adjust or stop.
After the pilot
What they keep, and what you measure
- The task map, even if they stop after the pilot
- A working setup with their saved rules and examples
- A review routine that fits their day
- Your honest go or no-go read
- Drain rating from 1 to 5, before and after
- How often a first draft goes out with only small edits
- Review minutes against how long the task used to take
Template
A scope note you can send
First step: one task, [task name]. Over four weeks I'll walk through how you do it today, set up an assistant with your rules and examples, and run it with you approving everything before it goes out. You keep the task map, the setup and the saved rules whether or not we continue. Fixed price: [amount]. At the end you'll get an honest read on what it saved and whether it's worth expanding.
Swap in the task and price before you send it.
The stop rule
If reviewing the output takes as long as doing it by hand, say so, then change the task or end the pilot.
Saying it early makes every other recommendation you give more believable, and that credibility is worth more than one pilot.
Session checklist
Before, during and after the walkthrough.
Thirteen steps. Each one you tick lights up a block in the model.
Tap the model · swipe
Before the session
Prepare
During the session
Walk it
After the session