Level 3 taught you to run AI-assisted work as an operation a small team can repeat. Level 4 asks you to govern it across the whole business, and to explain what you decided to a customer, an insurer, a lender, a franchisor, a buyer of the business or a state attorney general. It is written for owners and co-owners who carry final accountability for how the business uses AI, and for the general managers and operations directors who hold delegated authority for it, in businesses with roughly one to fifty staff and no counsel on retainer.
The first four modules place accountability and map what binds the business. Module one explains why accountability disappears when everyone does everything, what no tool, vendor, contractor or agency can take off the business, when the accountable person is a licensed professional rather than the owner, and how an owner with no compliance function carries it alone. Module two writes an AI policy short enough to be read and specific enough to be followed, sized for a team of thirty, three or none, borrowing a management system's plan-do-check-act shape without adopting the standard, and meeting the security basics the business already needs. Module three keeps the decision record: the choice, the alternatives and the trigger to revisit, what a new manager, franchisor, insurer, lender or buyer needs to see, and the payroll and employment eligibility records the law already requires at any size. Module four builds your own regulatory map on the size gate: federal headcount thresholds and what the next hire changes, four state privacy tests built four different ways, what "small business" means when the SBA defines it industry by industry, where federal AI policy stands as movement rather than settled law, and which entries only a professional can settle.
Modules five to seven cover what the business buys. Module five assesses an AI supplier in the time an owner actually has, tells an answer from an evasion, borrows a financial institution's service-provider requirements as a model, and decides what to do when a supplier will not answer and the business still needs the tool. Module six asks what a vendor's accuracy or savings figure was measured on, by whom and against what, borrows the method rather than the numbers from an independent evaluation, reads each study by its design, and runs a small acceptance test on work whose answers the business already knows. Module seven reads the data-use, training and subprocessor terms, plans for changes to the model, the price or the terms, gets data out and wipes what is retired, and decides whether to accept, narrow, negotiate or walk away.
Modules eight and nine handle security and fraud from the defender's side. Module eight takes the few controls that matter most from voluntary federal guidance written for owners: multi-factor sign-in, backups proven by a restore, updates and account recovery, where AI accounts and connections sit, ransomware planning, and New York's reasonable safeguards duty as an example of a duty that scales rather than exempts. Module nine describes synthetic voices, video and messages as the reporting bodies describe them, builds call-back on a number the business already held before any payment, bank change or reset, explains why a written rule beats judgement, and looks at the AI business opportunities sold to owners themselves.
Modules ten to twelve cover the people the business deals with. Module ten sorts every AI touchpoint into required, advisable or merely honest disclosure, compares three state disclosure duties built three different ways, sets out monitoring and recording notices for staff, and honours a request to deal with a person. Module eleven marks where this programme stops and an employment lawyer starts, covers hiring duties that do not wait for fifteen employees, impact monitoring with the four-fifths rule of thumb, screening companies and the payroll record a scheduling tool may become. Module twelve covers accessibility duties that ignore size and building age, what an overlay cannot promise, the customers a chat or voice agent serves badly, and a human route that is always there.
Modules thirteen and fourteen make the business answerable. Module thirteen gets AI-assisted work checked by someone who did not build it, borrows the shape of assurance without pretending it applies, and assembles the evidence pack before anyone asks. Module fourteen hears staff resistance before overruling it, announces and teaches a change that can be reversed, reads two adoption surveys that asked different questions, sequences the programme against cash and capacity, and prepares the capstone: an AI Governance Programme, Scenario Defence and Ownership Report for a fictional business.
Statements of authority say whom they bind, at what size and in which state; voluntary frameworks are never taught as law, and no framework is claimed to reduce harm. The level ships with a printable workbook and thirteen templates, and the examination draws forty scenario questions from a reviewed bank.
Everything here is professional education. It is not legal, tax, accounting, employment, privacy, security or insurance advice, and it does not replace an attorney, a tax professional, an insurance broker, a licensed professional where a trade or profession requires one, or the law of your state. Completing the level earns an independent educational certificate issued by AI Coalition Network with a public verification page. It is not a business licence, a professional qualification or a government approval, it carries no professional education hours, and it confers no authority to practise any licensed trade or profession.