It depends on what you want. Here is what it depends on.

Kaima

It depends on what you want. Here is what it depends on.

The most common question we are asked, close to word for word, is “how much is AI implementation going to cost me?” It is a fair question and it is the point at which someone has stopped browsing.

It depends on which of three quite different things you are buying. They are not sizes of the same thing. They are different jobs with different risks, and the gap between them is much larger than the gap between a small one and a big one of the same kind.

Level one costs least because it is bounded

One process that matters to you, working properly. Quoting, intake, the follow-up that keeps getting missed, the report somebody rebuilds by hand every month.

This is the lowest-risk thing to buy and the fastest to know whether it worked, because the process either runs or it does not. It is also where we are strongest: the front end of a business, where the work meets the customer.

What moves the number here: how many systems the process touches, and whether the process is actually agreed. A process that three people run three different ways has to be decided before it can be built, and that decision is yours rather than ours.

Level two is priced against a number you already track

Not a new process, but an existing one that has to get better against a number you already track. Response times, conversion at a particular stage, how much rework a team is absorbing.

Level three costs most, and starting there costs more still

Everything joined up, your own knowledge in the core of it, and visibility across the business rather than in six separate places. This is what people usually mean when they describe the problem as fragmentation: a stack of software that does not talk to itself and an owner who cannot see their own business without asking three people.

It is the largest of the three and the only one with a real obligation on your side: your knowledge has to go into it, and that is time from the people who hold it. A connected system built on generic content returns generic answers, confidently.

What moves the number here: how many systems, what condition the data is in, and how much of that knowledge is written down anywhere today.

The four things that actually move the price

Across all three levels, the same four things do most of the work. If you want to estimate roughly where you will land before you speak to anyone, these are the questions to ask yourself.

1. Which level you are actually buying. The largest single factor, and the one most often mis-set at the start, usually because the pain is described at level three and the sensible first engagement is at level one.
2. How many systems it has to touch. Each connection is a separate piece of work, and the cost is in the system on the other side rather than in the AI.
3. What state your data is in. Not whether you have a lot of it. Whether it is consistent enough to be relied on, and whether the same thing is called the same thing in two places.
4. How much of your own knowledge has to go in. Zero for a contained process fix. Substantial for a connected system, and that is time from your people, not a line item we can absorb.

What makes a project cost more than it should

Three things, in the order we see them.

• Building before the process is decided. If the business has not agreed how the work should run, software will not settle it. It will encode one person's version and make the disagreement more expensive to resolve.
• Treating safety and governance as a final step. Where data may sit and who may see it are structural decisions. Made late, they mean building it twice.
• Skipping adoption. A tool nobody uses costs everything and returns nothing, and it is the most common way a technically correct build produces no result at all.

Where we say not yet

We do not have a list of work we refuse. We do have a shape of engagement that is worth what it costs, and it is fairer to say where that stops than to let you find out on a call.

• Large enterprise is not what we are built for. Two founders on every engagement is the point of us, and it does not scale to a procurement process.
• Deep back-end systems work is possible but it is a larger undertaking than the front-end work we are strongest at, and we will tell you that before you commit rather than after.
• If the return does not clear the cost, we will say so. That is not modesty. A project that cannot pay for itself is one we would rather not have delivered.