Level 1 taught how AI systems work and fail on drawings, specifications, contracts and numbers, and whose duty each output is. Level 2 puts that to work on the tasks that fill a contractor's day. It is written for project managers and project engineers, estimators and preconstruction staff, superintendents and field supervisors, safety coordinators, contract administrators, office staff, and the working owners of small contracting businesses. Every module takes one task and builds a supervised, repeatable, defensible way to do it, with the reviewer's step and the qualified person who owns it named on the page.
Modules one to three build the instrument. Module one sets the operating model: triaging a task before any tool opens, running an assisted task from the project documents to a reviewed deliverable, reading a vendor accuracy claim for what was measured, on what and against what, and keeping the project AI use record while the work happens. Module two builds the reusable prompt: a project header naming the jurisdiction, delivery method and contract form, a manifest of documents at their revision, constraints and tie-outs, an instruction to refuse when the documents are silent, and defences against instruction-like text inside submittals, emails and scans. Module three turns saved prompts into a company prompt set with owners, versions, a test against a known answer, retest triggers such as a new code edition, and deliberate retirement.
Modules four to six are production work. Module four assembles the verified project fact file, writes closed-file drafting instructions, traces every statement in a draft back to its document, and holds the line between a first draft and a proposal, letter or notice a named person releases. Module five extracts specification, submittal, subcontract and purchase order requirements into a checklist a person verifies line by line, organises findings by the project's own classification conventions, compares terms against the prime contract, and says plainly what the checklist did not cover. Module six is the project record: what an AI-assisted document must show, the drawing and specification revision it was written against, a reviewer sign-off that states exactly what was checked, and retention for the periods the law already sets.
Modules seven and eight are research and verification. Module seven explains what a linked citation does and does not prove, how a grounded answer is still wrong on a real job (the wrong edition or jurisdiction, a local amendment, a voluntary standard read as binding), how to frame a question so those limits show, and what the measured ceiling on automated code checking means for the person who verifies. Module eight is the five-step Authority Verification Protocol: does the authority exist, does it say that, is it adopted here in this edition, is it in the executed documents, is it voluntary or binding, and how the check is recorded so a second person can repeat it.
Modules nine to twelve follow the work to the customer and the money. Module nine covers estimating work an estimator must own: a scope pulled from a specification section, quotes that each exclude something different, quantities tied to a sheet and revision, and what a public bid commits the bidder to. Module ten builds owner, homeowner and subcontractor messages the record supports, strips volunteered predictions and admissions, and handles delay and change messages that neither concede nor promise. Module eleven sets the handoff point for intake assistants, runs automated calls and texts on recorded consent, and checks residential contracts, cancellation rights and licence numbers against California, New York, New York City and Maryland rules taught as those places' rules. Module twelve keeps invoices, proposals and payment applications truthful when AI shortens a task, passes tool costs through honestly, explains the certifications a person signs on covered public work, and states savings only from the company's own measured baseline.
Modules thirteen and fourteen close the loop. Module thirteen classifies project material at the moment it is about to enter a tool, separates the customer and worker data duties that depend on size thresholds from those that do not, and covers biometric time clocks, recorded calls and AI note-takers, financing and card data, and the confidentiality that follows project data into a third-party tool. Module fourteen tests what the project AI use record must answer and for whom, works the first hour after an AI or data incident on a project, explains the breach notice, Safeguards Rule and OSHA reporting clocks that already bind a contractor, runs the review afterwards, and assembles the Level 2 portfolio.
Statements of authority are labelled with whom each binds and when it was checked; no state's rule is presented as a national one. The level ships with a printable workbook and nine templates, a final examination that draws forty scenario questions from a reviewed bank, and a capstone Project Workflow Portfolio built on fictional companies, projects and customers.
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. Completing the level earns an independent educational certificate issued by AI Coalition Network with a public verification page. It is not a contractor licence, a trade or safety credential, a competent-person designation or a compliance certification, it carries no professional education hours, and it satisfies no licensing, registration or safety-training requirement.