Level 2 taught one person to do one business task well with AI, with the owner's review and a record. Level 3 is about the operation: the same work run by several people across many customers, where the owner can no longer read every quote, reply and confirmation before it goes out. It is written for owners, co-owners, general managers and operations leads of businesses with roughly one to fifty staff who already use AI every day, including a sole owner getting ready to hand work to a first employee, a seasonal helper or a virtual assistant.
Modules one to four build the operating frame. Module one explains why a task that was safe when the owner did it once becomes risky when a helper does it forty times, why a good week breeds complacency and checking everything equally, how to name who owns an operation when everyone does everything, and why the business captures its own baseline before any redesign. Module two walks the job as it actually runs from first enquiry to paid invoice, records the undocumented steps, handoffs, waiting and rework, counts the baseline, and shows why measured and reported hours disagree. Module three builds a four-tier grid on consequence and reversibility, attaches check depth, checker, record and escalation to each tier, and names what pushes a task up a tier however small the volume. Module four writes a procedure that opens with its hard stops, sets out the trigger, data class, tool, prompt entry, steps, check, escalation, record, measures and retest triggers, tests it by watching a newcomer use it, and borrows a plan-do-check-act rhythm without adopting a management standard.
Modules five to seven govern the knowledge and the data. Module five chooses the price lists, policies, warranties, service areas, templates and past answers worth including in a knowledge base, tags each by date, state, service line and version, gives every document an owner and a review date, and keeps card data, biometrics, health records and employee files out. Module six shows how a tool searching the business's own files still gives a customer the wrong answer, through a superseded price list, an expired promotion, another shop's hours, another state's terms, the wrong company or a competitor's promise, and answers with the citation rule, the answer boundary and a fixed test set re-run after every change. Module seven applies least privilege inside tools built for sharing, removes access when contractors and seasonal staff leave, sets multi-factor sign-in and logging without a second copy, and explains which businesses the FTC Safeguards Rule reaches and what stays on below five thousand records.
Modules eight to ten put the frame to work at volume. Module eight handles enquiries across phone, web form, text and social media, sends anything unclassifiable to a person, names the intake triggers for safety, money, a legal threat and a regulated question, keeps first replies to promises the business can keep, and discloses the automated responder under three different state positions. Module nine quotes from one controlled rate sheet, recomputes and ties out a day's quotes before any figure leaves, sets out the consent position behind scheduling, dispatch and confirmation messages, and covers the notice questions an AI notetaker raises on a call. Module ten explains what changes when a business sells beyond its own state: economic nexus and why each state sets its own terms, shown through California and Texas, tracking sales by destination, licences when work crosses a state line, marketplace duties toward high-volume sellers, and using AI to prepare a question for a tax professional.
Modules eleven to fourteen keep the operation honest over time. Module eleven builds a watchlist of the official sources that move for the business, with an owner, a frequency and a check log, names the places a rule can stand, from proposed and delayed to set aside or in force, on dated examples, and carries one confirmed change through the knowledge base, the procedures and the saved prompts. Module twelve replaces "it seems fine" with a sample checked by someone other than the person who produced the work, sets how much to sample and when to sample more, and uses a seven-code error list with thresholds that trigger action. Module thirteen agrees escalation paths and stop conditions before they are needed, keeps a manual route back that someone still knows, and starts any system that acts rather than suggests with low-risk uses. Module fourteen chooses measures that would reveal failure against the business's own baseline, shows how each gets gamed, decides on the evidence to narrow, pause or switch off, and prepares the capstone.
Statements of authority say whom they bind, at what size, and when they were checked; a duty with a size gate is labelled with it, no state's rule is presented as a national one, and voluntary frameworks are never taught as law. The level ships with a printable workbook and thirteen templates, a final examination that draws forty scenario questions from a reviewed bank, and a capstone Workflow Redesign of one workflow built on fictional businesses, customers and figures.
Everything here is professional education. It is not legal, tax, accounting, employment, privacy or licensing advice, and it does not replace an attorney, a tax professional, a licensing board, 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.