How we decide

We decide where it pays before we build a thing.

Most of the work isn't building. It's knowing where AI pays.

Plenty of people will sell you software. Few will sit down and show you where it earns its keep in your own week. That's our starting point.

Every department has some.

Dull, repeated work hides in every corner of a firm.

• Sales: promises in the sent folder, and enquiries left until tomorrow.
• Finance: late payers chased at month end, and receipts hunted down.
• Operations: the owner's inbox, and approvals stuck on one desk.
• HR: new starters with no login, and leave requests that clash.
• Customer service: forms that bounce, and documents nobody sends.
• Marketing: clients who were never told what else you do.

It looks much the same in staffing, accounting and insurance. Wealth management, home health and manufacturing have it too.

We follow the work, not the org chart.

We trace how a job really moves. Who asks. Where it sits. Who chases it.

We walk a few chains end to end. A new client to a first invoice. An offer to a first shift. An enquiry to a reply.

Each chain stalls somewhere. Usually a person is reading, sorting or chasing by hand. That's where we look.

We look at the owner too. Promises, decisions and relationships live in your head. In a heavy week, some slip.

Where it does well

It's good at work that is mostly reading and follows rules. Jobs like these:

• going through a full inbox and pulling out the few that need a person;
• spotting a promise in a sent email and keeping track of its date;
• checking what came back against what was asked for;
• drafting a first reminder for someone to approve;
• noticing when a pattern changes, and saying "worth a look".

A sharp assistant could do each of these with enough hours. Nobody has the hours. That's where it pays.

Where it doesn't

• calls that need judgment you'd never hand over;
• work that already runs fine;
• gaps that are really a missing rule or a missing hire;
• jobs too small to be worth building for;
• records too messy to read well yet.

We'll tell you when AI is the wrong answer. Often the fix is a checklist, a named owner or a tool you already have. We'd rather say that than sell you something you don't need.

What we build reads, sorts and drafts.

It never acts in your name. A person approves every message before it goes.

When it isn't sure, it says "worth a look" and leaves the call to you.

The first conversation

It's with one of our co-founders. It's free and runs about an hour.

• No access needed. We won't ask for logins or mail.
• Nothing to prepare. Just turn up.
• Bring the whole list. Whatever brought you here is one item. Bring the others.
• We ask about your week. What waits, who chases, what lands on you late.
• You leave with a short list. What's worth doing, in plain words, and what isn't.

No sales pitch at the end. If nothing is worth building, you'll hear that. If something is, you choose what happens next.

Where you end up

You spend on AI only where it counts. Your people spend their week on the work that needs them.