A Danish consulting firm

Do twice as many clients have to mean twice as many people?

In consulting the answer has usually been yes. You sell people's time and judgment, the calendar fills up, you hire the next good person, and then you can take on more work until that calendar is full too.

There is nothing wrong with running a business that way. It does mean headcount sets the speed limit on growth.

We are working with a small Danish consulting firm on changing that.

The case is anonymized, so we won't describe the firm, its clients, or the service itself.

Not fewer people. More capacity per person.

The goal was never a consulting business without consultants. Clients pay for exactly the part that takes experience, craft, and judgment.

But a working day is not only made of those moments. There is a lot of work before, between, and after them: finding the information, reading through the material, getting it into an order that holds. Digging up what the firm did the last time a case looked like this. Keeping the documentation current, carrying new details forward through the case, building the deliverable and then checking it and fixing what is wrong.

That work is necessary. It just doesn't need an expert hour every time.

So we built around the expert

The Coworker now works alongside the consultant across the whole client engagement, not just one task at the start or one document at the end. It holds the context as the case moves, does parts of the work itself, and sets up the next step with what it has already learned.

When something calls for professional judgment, the consultant takes it. When something follows from rules, context, and work that are already in place, the Coworker takes more of it.

Where that line sits is the part that matters. We are not trying to make AI look like a consultant. We are trying to spend the consultant's hours on the parts that need a consultant.

It changes how the firm can grow

Say a service takes 100 hours of human work to deliver. 10 clients is 1,000 hours. 20 is 2,000. Sooner or later the answer is obvious: hire.

But if a Coworker can carry part of those 100 hours, the arithmetic stops being a straight line. The next hires don't disappear. Revenue, client count, and headcount just stop having to move at the same speed. A firm with 5 strong consultants might handle work that used to take 7, and when the 6th one starts, capacity goes up again.

That is the difference we are after. Not zero headcount, just headcount that no longer has to grow in step with the work.

"Big consultancies have always had one advantage: hands. That advantage is worth less when a good consultant can get far more out of their own."

Sewar Sidou, Ellypsis

Something else comes with it

A Coworker can only support a service properly if it can find its way around the work: what has happened, where a piece of information came from, what has already been decided, what is still open, and what the next step builds on.

So structure grows around the work almost as a side effect. Knowledge that used to live in an inbox, an old document, or the head of whoever ran the case becomes something the whole firm can work from. That is worth having even if you took the AI back out.

The next hire starts somewhere else

This matters more as the firm grows. The next consultant still has to learn the craft. No Coworker can install 10 years of experience in a person.

What surrounds the craft is a different story. It no longer has to be picked up by slowly working out how this particular firm does things. The work already has a structure, earlier cases leave something usable behind, and it is clearer where information comes from, how the pieces fit together, and what is expected along the way.

So the Coworker gives the current consultants more capacity, and it also makes the firm less dependent on knowledge that only lives with one or two people.

It had to work on real clients

We didn't finish the system before it met a client. It went into real work and grew from there, and that has meant corrections.

Parts that looked right on paper were not good enough in use. Some steps fired too early. Others dropped important information on the way through. One piece worked fine for one person and badly for the next. So we fixed them.

The Coworker covers the whole engagement today, and it still changes, because the work changes and because real use finds things a workshop never will.

This is the real test

Whether AI can produce a document faster is not a very interesting question. It can.

The interesting question is what happens to a firm when expert work stops being the same thing as human hours all the way through. Can 5 people deliver what used to take 7? Does the next hire get into the work faster? Does the firm keep its own way of working as more people join? Can it grow without the organization having to grow just as fast?

That is what this Coworker is built to answer. Not a consulting business without consultants, but one where people are no longer the only way to add capacity.