Enterprises get an AI governance committee. A nine-person company gets a subscription and whatever the default configuration happens to be.
Small businesses now run meaningful decisions through AI — who to call back, which invoice looks wrong, how to answer a customer, what a review really said. The tools assume an organization with a compliance department, a policy team, and time to review dashboards. A small business has none of those.
What it has is an owner with fifteen minutes a week and a default configuration nobody chose deliberately. Every governance framework aimed at this problem is written for the company that already has staff to run it, which is precisely the company that isn’t the small business. The gap is not awareness — it is that responsible oversight currently costs more attention than the business can spend.
Project Keel asks what Corvion’s emotional-governance and causal-influence research look like when the entire oversight budget is one non-expert person for a few minutes a week. It treats attention as the scarce resource the design has to respect.
Studying whether the safe setting can be the one that ships, rather than one an owner has to assemble.
Surfacing the few things that actually need a human, instead of a dashboard nobody opens.
Asking whether the system can notice when its own behavior has drifted, without an analyst watching it.
A plain, exportable account of what was decided — for a customer, an insurer, or a regulator who asks later.
Project Keel studies how AI oversight can fit the constraints of a small business. It does not certify, audit, or attest that any business is compliant with any law or standard, and it is not legal, accounting, or regulatory advice.
This is a live research direction at Corvion, run in the open by design. The work centers on a single question: what governance looks like when the reviewer is the owner, and the review budget is fifteen minutes a week.
Our approach is to define that question rigorously — the criteria, the failure modes, and what would count as an answer — before building anything to serve it. We publish findings here as the work develops, including the results that don’t hold up.
We’re looking for small-business software vendors, managed service providers, industry associations, small-business development centers, and owners willing to tell us what actually fits a fifteen-minute week.