This is the foundation level of the Master AI General Contractor Certification, rebuilt as a master class. It is written for the people who build things and run the businesses that build them: contractors and remodellers, project managers and superintendents, estimators and schedulers, safety, quality, procurement and office staff, and the owners who set company policy. It assumes you know construction and nothing about artificial intelligence. A learner who finishes can decide, for one real task on one real job, whether and how AI may be used, who owns the output, and how they would know if it stopped working.
The first two modules take apart what is being sold to contractors as "AI". Module one separates rules-based automation, machine learning, computer vision and generative AI, shows how a language model assembles an answer one token at a time, opens up the retrieval, tool use and agents inside construction software, and surveys the tool categories. Module two is about failure: an invented specification section, a code answer written for an edition the jurisdiction never adopted, a quantity that does not tie back to the takeoff, a clause read out of a contract nobody executed — each with a fictional example you can check line by line — and the automation bias that lets a plausible draft through on a bid day.
Modules three to five cover the duties that do not move when a tool helps. Module three classifies what a contractor holds — owner drawings under nondisclosure, subcontractor pricing, employee records, homeowner payment details, biometric time data, site recordings — shows where consumer and business terms each put it, and ends with a five-step decision on whether an item may enter a given system. Module four maps accountability: who is answerable for each output, five genuinely different state licensing models so you never mistake one state's rule for the national rule, and the sealed-design line a contractor may not cross. Module five builds a research workflow running from a well-framed question to a verified answer at the publisher, and explains why a model code binds nothing until a jurisdiction adopts it.
Modules six and seven are about producing work. Module six builds the ten-part construction prompt, insists that you supply the drawing, specification or contract rather than asking a model to recall it, and gives patterns for six tasks a contractor repeats every week. Module seven takes a first draft of a proposal, RFI, change-event narrative, daily report or owner letter and shows how to get it without inheriting its first mistakes: a verified project fact file, a hunt for the invented date, quantity and clause reference, a separate treatment for documents whose contents a statute fixes, and a record of who checked.
Modules eight to ten follow the work through the business. Module eight covers preconstruction — pulling scope out of specification sections, levelling quotes, drafting proposal narrative — and the rule that no quantity, price or total leaves the office unless a person can open the document it came from. Module nine is safety: the named duty-holders a safety programme is written to, the training a person delivers and signs, and the inspections made on the face of the work. Module ten is the office and customer side — intake, scheduling, back-office extraction, reviews, advertising claims, calls, texts and AI-generated voice — ranked by the harm a wrong output would do.
Modules eleven to fourteen build control. Module eleven turns ad hoc use into a staged workflow — inputs, data classification, the AI step, verification, reviewer sign-off, record — designed so a wrong output cannot become a wrong action on site. Module twelve bands agent uses from read-only summarising up to sending a notice or moving money, and fits control points, stop switches, permission limits and separation of duties to each band, including instructions hidden in documents the company receives. Module thirteen covers vendors and security: what a vendor actually promises, who owns the project data and the model output, and running identity, access, retention, disposal and exit like a contractor rather than an IT department. Module fourteen writes the company AI policy where no regulator has said what it must contain, and fits AI into the quality and safety programmes already in place. A printable workbook and ten templates ship with the level.
Module fifteen is a practical lab of four exercises on invented projects: find the planted errors in an AI-drafted scope and change-event narrative; catch the invented specification section and the citation to a code edition nobody adopted; review a submittal extraction and design the workflow around it; then evaluate a vendor, build a policy section and check marketing copy. Module sixteen is the capstone, a Construction AI Implementation Plan for one low-risk first use, with its data classification, controls, named owners and reviewers, vendor evidence, baseline measures, failure indicators and stop authority. It promises no savings, because no construction benchmark here supports one.
Everything here is professional education. It is not engineering, design, legal or safety advice; it does not teach code compliance determination, engineered design, or the means and methods a competent person decides on site; and it does not replace a licensed design professional's seal, the authority having jurisdiction, a competent person on the ground, legal counsel, or the executed contract and the law of the state where the work is performed. Every statement of authority is labelled with whom it binds and when it was checked. Completing the level earns an independent educational certificate from AI Coalition Network with a public verification page. It is not a contractor licence, a trade or safety credential or a compliance certification, and it satisfies no licensing, registration or safety-training requirement.