Start with the job that wastes the most time

Every company we talk to already has a list. Usually it is long, and usually the first thing on it is the most exciting idea someone had in a meeting. A chatbot for customers, a dashboard for the board, something that sounds good in a pitch.

We park that list for a moment and ask a more boring question: where do the hours actually go?

The question behind the first agent

An agent is software that runs a whole process end to end, for example chasing invoices, entering orders or checking documents. It hands over to a person whenever a decision needs a human. So the first agent should sit on a process that eats a lot of time and that someone is still doing by hand today.

Most SMEs and mid-market companies run a big part of their day on email, spreadsheets and copy and paste. Nobody chose that. It grew one workaround at a time, and after a few years it is simply how things get done. Those jobs rarely make the list, because nobody sees them as a project. They are just Tuesday.

Frequency × Effort

The tool we use to pick is simple enough to draw on a napkin. Two axes: how often a job happens, and how long it takes each time. The hours are basically one times the other.

A job that happens once a quarter and takes a day is annoying, but it is rarely where the hours go. A job that takes a few minutes and happens all day long adds up fast, and nobody notices because it never shows up as one big block of time.

The top right corner is the one we care about: jobs that happen often and take long. Orders that arrive by email and get typed into another system. Claim documents that someone reads line by line. Invoices that need chasing every week. That corner is where the first agent goes.

Why the exciting idea waits

The ideas on the original list are often good. They just tend to sit bottom left: things that happen rarely, or take little time when they do. Building them first means spending the effort where the return is smallest.

There is a practical reason too. The first agent is how a team learns to work with agents. If it takes a job everyone hated doing, people feel the difference in their own week, and the conversation about the second one gets a lot easier.

How we find the corner

You can't plot what you haven't seen. This is why every engagement starts with a paid assessment: our engineers sit with your team, follow the work and mark where the hours go. We call the result The Map. Every job on it gets a place on the Frequency × Effort grid, and the top corner usually becomes obvious.

Usually, because sometimes two or three jobs sit close together up there. Then we look at what else each one touches, and at where the hand-over to a person would sit. The one with the cleaner line tends to go first.

Then the next one

Once the first agent is running, we go back to the map. The Coreform OS keeps it up to date as we learn, and over time it shows where the next agent should go. One process at a time, each time starting from the corner that costs you the most hours.

Each agent also makes the next one faster to build, because agents we have built before are adapted for your company instead of starting from scratch. So the order matters: the first agent takes back the most hours, and it makes the ones after it easier.

If you want to know where your corner is, book a call with one of our founders, or start an assessment.