Doctrine: how this firm works
Fractional AI Advisory is a fractional AI advisory firm for owner-led businesses in the Des Moines metro, with revenue under $10 million. This page states how the firm works and what it holds to. The rules below are the firm's public position. These rules are written into the firm's engagement terms, not left as slogans.
1. Measured results only
Every engagement starts with a baseline. The baseline is measured before any build begins: what the workflow costs today in staff hours, and what it costs in errors and rework. Results are reported against that baseline, with the method stated. If a number cannot be traced to a baseline, it is not claimed. Reports also state what the numbers do not prove. There are no invented savings in this firm's reporting, and no savings figure is published without the baseline it came from.
2. Guardrails are a public position
This firm advocates for guardrails on AI use in public, and it works under them in its own engagements. Guardrails are not a feature of a package. They are the condition the work runs under. Every engagement works under a written AI policy: which tools are approved, what data may go into them, and who approves a new use. If a client has no policy when work starts, adopting one is part of the first engagement. If a tool or a workflow cannot run under the policy, the firm says so in writing and recommends against it.
3. The human gate
AI does not send, post, pay, or decide on its own. Anything that leaves a business, moves money, or commits a business to a third party passes a named human first. The gate applies to this firm's own work product as well. Work the firm delivers is reviewed by a person against its source material before delivery. AI output that a client's customers will touch is held to the review controls in the client's policy, and client-facing builds get an adversarial review before launch: a reviewer whose job is to break the build runs it against failure cases, and the findings are fixed or accepted in writing first.
4. Adoption is the work
A workflow nobody uses returns nothing. Installing a tool is not the result. The result is a team doing its repeat work a new way, and keeping at it. This is why the firm's reports count adoption rate: how much of the eligible work actually ran through what was built. It is why training repeats on a cadence instead of happening once. A build with a good measurement and a low adoption rate is reported as what it is, a build the business has not taken up yet.
5. One workflow at a time, run as a pilot
Builds are not programs. Each build is one workflow, run as a pilot with the discipline stated in advance. There is one owner on the client's side, named in writing, and one problem, stated in one sentence. There is one measurable expectation: what will change, by how much, and how it is measured. The baseline is measured before the build starts. The time box and the scale-or-stop decision are agreed in advance, and the decision is made on the measured result. A second process is a second build. The order of builds is set by a plan, and the plan says what to skip as plainly as what to fund.
6. Advisor, never officer
No one from this firm is an officer, employee, or agent of a client's business. The firm advises, plans, and builds. Decisions stay with the owner. A client who wants the outcome of an AI lead without hiring one gets the work of one, part time. The authority stays where it belongs.
7. Local, and sized to the client
The firm's home market is owner-led businesses under $10 million in revenue in the Des Moines metro. The work is sized to that market: engagements priced in the open, one workflow at a time, and no full-time hire required to own the results.
Book a call
Email david@fractionalaiadvisory.com. Fractional AI Advisory, Des Moines, Iowa. Fractional AI Advisory LLC is an Iowa limited liability company.