Advanced Construction Skills AI capability guide
Format
Standalone guide

Codex 101 for Construction PMs and VET Leaders

Codex is not just a chat box. It works inside a project folder on your computer, where the real work already lives: RFIs, change orders, site photos, modules, mapping tables, learner evidence, templates, and review notes.

The Key Idea

Codex is useful because it works inside a project folder on your computer. It can read files, compare them, create new ones, and follow standing instructions across sessions. You do not need to code to use it.

Quantity surveyor evidence workflow visual
A familiar scenario first: Codex helps make sense of a folder full of RFIs, instructions, quotes, photos, and claims.
Flip
Quantity surveyor scenario

Load RFIs, change orders, site instructions, subcontractor quotes, daywork sheets, photos, claims, email threads, and the contract schedule. Ask Codex to trace what changed, what it affected, what evidence supports it, and where the cost impact may sit.

What Codex is doing

It is building an evidence trail: event, source documents, dates, cost category, missing records, assumptions, and items needing QS review.

Human still owns: entitlement, valuation, contract position, and anything submitted commercially.

Flip
VET compliance scenario

Load units of competency, assessment tools, mapping documents, TAS files, learner samples, validation records, trainer credentials, and existing modules. Ask Codex to map coverage, identify gaps, flag duplicate coverage, and prepare a review pack.

What Codex is doing

It is building a coverage map: requirement, module location, assessment task, evidence source, weak coverage, duplicate coverage, and review questions.

Human still owns: compliance judgement, assessment decisions, validation, and RTO accountability.

Professional boundary

Codex can prepare the evidence trail. It does not own entitlement, valuation, safety sign-off, assessment judgement, compliance approval, or trainer competency decisions.

Part 1: What Is Codex?

Codex is a tool you install on your computer, the same way you would install Zoom or Zotero. Once installed, you open a project folder and let Codex work inside it. That folder might be a dissertation, a paper, a job folder, a training program, or a compliance review pack.

Inside that folder, Codex can read the files already there, edit existing files when you allow it, and create new ones. It also picks up saved sessions and project instructions each time you return, so you do not have to explain the same basic context every time.

Codex project folder concept screen
Codex starts to make sense when the user sees the folder as the workspace, not the chat box.

This is the key distinction from browser-based tools like ChatGPT or Gemini. Those tools mainly talk with you in the browser. Codex can act inside your files and folder, with the permissions you give it.

That matters whether you are an academic managing a corpus of PDFs and interview transcripts, or a practitioner managing variation registers, SWMS documents, and apprentice logbooks. The underlying capability is the same: Codex reads what is in your folder, works across multiple files, and produces outputs you can actually use.

Codex folder-based work output concept screen
Codex becomes useful when it can read the folder, compare files, and produce something you can review.
Plain-English version

You bring the AI to the work folder instead of copying pieces of the work into a chat window.

Part 2: Installing Codex App, Your First Session

Keep this bit calm and small. These are the three screens most new users need to recognise before they can start useful folder-based work.

Flip
Codex first launch screen with Playground selected
1. First launch: sign in and start in Playground
What this means

Nothing technical is happening yet. Codex is just open and ready. Playground is a practice space, not the project folder you will use for real work.

Next move: Add one safe practice folder before touching live job or compliance files.

Flip
Codex add new project screen beside a folder list
2. Add project: choose the folder where the work lives
What this means

A project is just a folder. For training, use a harmless folder with sample PDFs, notes, or templates. For real work, choose the job pack or qualification review folder.

Next move: Check the folder contents before asking Codex to read everything.

Flip
Codex session showing a prompt and permission check before creating a file
3. First session: ask, review, then approve
What this means

The permission prompt is a safety feature. It lets the user approve file creation or changes before Codex writes to disk.

Next move: Start with read-only summaries, then allow file creation only when the proposed output makes sense.

Screen audit note

For first-time learners, keep screenshots to one job per screen: open Codex, choose folder, approve safe output. Avoid terminal-heavy screens, hidden settings, or code unless the learner needs them for the task.

The next section shows how users move from simple prompts to folder work, evidence mapping, reusable workflows, and governed routines.

Continue to levels
1
Level 1

The Prompt User

You open Codex and ask isolated questions.

This is useful for quick drafting, summarising, and plain-English explanation. A PM might ask for a clearer email. A trainer might ask for a learner instruction to be simplified.

QS or PM use

Draft a response about a delay or explain a contract clause in simpler language.

VET use

Rewrite learner-facing instructions or summarise a policy note.

Why people stall

Every prompt starts from zero. Codex has no access to the folder, project history, template, or evidence.

Move up

Open the actual project folder and give Codex a real task with context.

2
Level 2

The Folder Worker

Codex can now see the job pack, not just your question.

Give Codex the relevant folder and a clear definition of done. Tell it what to read, what to ignore, what output you want, and what must be flagged for review.

RoleFolder taskExpected result
QSRead RFIs, site instructions, quotes, photos, claims, and the contract schedule.Draft change-event register with document references and missing evidence.
VET complianceRead units, modules, assessment tools, learner evidence, mapping documents, and validation notes.Draft coverage map and compliance questions for review.
Why people stall

They dump files into a folder and hope. Codex still needs instructions, boundaries, and output rules.

Move up

Ask Codex to trace every claim back to source documents. Evidence mapping is where the value starts to compound.

3
Level 3

The Evidence Mapper

Codex is now comparing, tracing, grouping, and showing its working.

At this level, Codex is no longer just drafting. It links claims to source documents, shows gaps, and separates evidence from assumptions.

QS mapping

Link each potential change event to RFIs, instructions, quotes, photos, dates, delay notes, cost categories, missing records, and items requiring QS review.

VET mapping

Link each requirement to module content, assessment tasks, learner evidence, validation notes, weak coverage, duplicate coverage, and gaps.

Good output includes

Source references, confidence levels, missing evidence, contradictions, assumptions, and items that require human decision.

Quality test

The table is only useful if it leads back to the evidence.

Why people stall

They accept the table without checking the trace.

Move up

Create standing instructions and turn the mapping task into a repeatable workflow.

4
Level 4

The Workflow Builder

You stop explaining the same review method every time.

AGENTS.md becomes your standing project brief. Skills become reusable workflows for common jobs.

QS skills

RFI-to-change-order impact review, variation evidence pack builder, and progress claim support checker.

VET skills

Unit-to-module compliance mapper, assessment evidence gap finder, and validation pack reviewer.

Reusable method

A good Skill defines the input folder, evidence rules, output table, review flags, file naming, and human sign-off point.

Why people stall

They get one good result but never save the process. The next review starts from memory again.

Move up

Choose one low-risk recurring review and make it routine, with a clear human approval point.

5
Level 5

The Governed Operator

Codex becomes part of a controlled review system.

The question is not, "Can Codex draft this?" The question is, "What is Codex allowed to read, what can it draft, what must it never change, who reviews the output, and what audit trail do we keep?"

QS routine

Prepare a draft change-event review pack from approved folders, list missing evidence, highlight commercial exceptions, and wait for QS sign-off before anything is issued.

VET routine

Check qualification folders for mapping gaps, evidence issues, validation records, version drift, and trainer currency notes, then produce an internal review pack.

Governance controls

Approvals, sandboxing, controlled connectors, Skills, automations, role-based access, and review logs keep the workflow reviewable.

Pattern example

The public Qualtrics-Codex starter workflow shows the same maturity pattern in another domain: secrets stay local, live system actions require approval, private raw data stays out of public outputs, and Codex works from prompts, scripts, docs, and project instructions to create a repeatable workflow.

Why people stall

Trust. The tool is not the hard part. The hard part is deciding where human accountability sits.

Best first routine

Start with one valuable but low-risk review. Keep the first version boring, reviewable, and human-approved.

Reference Summary

Use this view as a printable or LMS-friendly summary of the Codex maturity path.

Level 1: Prompt User

  • Use for isolated drafting, summarising, and plain-English explanation.
  • Move up when the task needs project files, templates, evidence, or history.

Level 2: Folder Worker

  • Open the project folder and define the output.
  • For QS work, create a draft change-event register.
  • For VET work, create a draft coverage map and review questions.

Level 3: Evidence Mapper

  • Require source document references, missing evidence, assumptions, contradictions, and human decision points.
  • Do not accept a polished table unless it traces back to evidence.

Level 4: Workflow Builder

  • Use AGENTS.md for standing project instructions.
  • Create Skills for repeatable QS and VET review workflows.

Level 5: Governed Operator

  • Define what Codex may read, draft, change, and never touch.
  • Keep approval gates, logs, and human accountability explicit.
  • Use the Qualtrics-Codex starter pattern as a model: keep secrets local, require approval before live actions, separate private inputs from public outputs, and package the workflow as reusable prompts, docs, and scripts.

Source Grounding

This guide keeps Codex claims tied to public OpenAI descriptions, then translates the examples into construction and VET contexts.

  • OpenAI Codex docs: Codex is described as an agent that can write, understand, review, debug, and automate development work.
  • OpenAI Codex product page: OpenAI describes multi-agent workflows, cloud environments, worktrees, Skills, Automations, and review/testing.
  • OpenAI Help: Codex surfaces include app, CLI, IDE extension, web, and cloud-backed workflows depending on plan and setup.
  • Codex best practices: OpenAI emphasises goal, context, constraints, done criteria, AGENTS.md, MCP, skills, and automations.
  • Codex enterprise guide: OpenAI covers local/cloud rollout, RBAC, managed policies, sandbox and approval controls, analytics, compliance, and governance.
  • Codex 101 for Academic Researchers: the ACS draft source for the folder-based explanation, adapted here for construction PM and VET leader examples.
  • Qualtrics-Codex starter workflow: a public example of a governed Codex project pattern with local secrets, approval gates for live actions, private raw data boundaries, and repeatable prompts/scripts/docs.