A hands-on training app for real-estate professionals who want to move from casual ChatGPT use into Codex workflows, reusable skills, batch listing improvement, approval dashboards, and self-improving quality systems.
A local real-estate folder with sample data, examples, rubrics, reports, and rules.
A reusable listing-improver skill that can be tested and improved.
A process for improving 5, 10, and 50 listings with review stops.
An approval dashboard so humans stay in control before publishing.
Each mission has exact steps, a prompt to paste, expected results, troubleshooting, notes, and a practical output.
Most people use AI as a clever answer box. In this mission, the learner experiences the difference between asking for help and delegating a small piece of work with a result, quality bar, and next action.
Time: 35 minutes
Agents are most useful when the work repeats. This mission turns a vague pile of tasks into a ranked list of workflow candidates.
Time: 45 minutes
Now the learner moves from conversation into a workspace. Codex can inspect files, produce reports, and create artifacts in folders.
Time: 60 minutes
The first useful win is improving one weak listing and explaining why the rewrite is better. The point is not just nicer copy. The point is evidence-backed improvement.
Time: 70 minutes
AGENTS.md teaches Codex durable project rules. In real estate, those rules protect accuracy, privacy, brand voice, and approval boundaries.
Time: 60 minutes
A skill is how the learner stops rewriting the same prompt. The starter workspace already includes a first skill draft for listing improvement.
Time: 90 minutes
This is where the learner feels the leverage. The same workflow that improves one listing can evaluate and rewrite many, but only if the output is structured for review.
Time: 90 minutes
Batch work needs a review interface. The learner should approve, reject, or request changes instead of hunting through a giant markdown file.
Time: 90 minutes
A harness lets the learner stop trusting vibes. It checks whether listing outputs meet a quality bar and catches repeated mistakes.
Time: 90 minutes
The final mission teaches the loop: when the learner corrects the agent, the correction becomes a better example, rubric, skill, or harness check.
Time: 90 minutes