AI capability installation, explained

AI adoption consulting sells judgment: an assessment of where AI fits your business, a roadmap, usually some training. It can be exactly right and still change nothing, because the model ends where your capacity begins. You get the slides. You still have to do the building, and most teams that had spare building capacity would not have hired a consultant.

Installation keeps the diagnosis but moves the finish line. The engagement does not end when the recommendation is delivered; it ends when the recommended thing is running inside your operation, staffed by your people, documented in your files. A consultant leaves you knowing what to do. An installer leaves you already doing it.

Neither is wrong. They answer different questions. Advice is the right purchase when execution capacity already exists in the building. It usually does not, which is why so many AI strategy documents are current, correct, and unused.

An AI implementation project builds something real: a chatbot, an automation, an integration. The builder arrives with a stack and templates, deploys, and moves to the next client. The system works on handover day. The open question is what happens after — who tunes it when the process changes, who understands why it answers the way it does, what happens when the person who championed it moves on.

The difference with installation starts with the raw material. An implementation is built mostly from the builder’s stack; an installation is built from your files — how the work gets done here, the decisions already made, the exceptions that make your business yours. That written context is what makes the result specific to you rather than a template with your logo on it. The discipline behind that layer is context engineering.

The second difference is who can run it. Installation treats training your operators and writing the governance as build work, not aftercare, because a system only your vendor understands is a dependency, not a capability.

A short test: name what you want to be true ninety days after the engagement ends.

• If the answer is we know what to do, buy adoption consulting. You need judgment, and you have hands.
• If the answer is a specific system exists, buy an implementation project. You need the build, and you accept depending on the builder for changes.
• If the answer is our team runs this without anyone’s permission, that is capability installation. Built from your files, owned by you, run with you until your people run it alone.

The observation underneath this page: the deliverable that matters is what your team still runs after the last invoice. Every other difference between these three offers follows from that one. And if you already bought AI and it went quiet, that is a different diagnosis — why AI rollouts stall covers it.

br8n publishes its numbers at the individual scale. The Working Intelligence Audit — the diagnostic that maps where your knowledge lives and which process to build on first — runs $397 to $997: $397 self-guided, $497 with a written review, $997 run live with Michael. A Personal Brain build is $1,497, or $4,997 for the deep version. Company installs start at $15,000 plus a retainer and run two to six months, priced after a diagnostic, because the honest number depends on how many workflows and people the capability has to reach.

The audit is the deliberate first step at any scale. It is bounded, it has a fixed price, and it produces the map — where knowledge lives, what repeats, what to build first — that any of the three paths above would need anyway. You keep the map either way, whoever you hire next, including nobody.