Home / AI Courses / Content on Autopilot / Full curriculum
Full curriculumAll sales are final, so everything you would be buying is listed here first. Every lesson below is a single specific outcome. You get the exact prompt to copy, how to use it, what goes wrong, and one thing to do on your own work before moving on.
Auto pilot is not a tool you buy. It is a chain: an idea becomes a script, the script becomes a voice, the voice becomes a timeline, the timeline pulls in images, and a file comes out the other end without you touching a video editor. This module builds that chain on your own computer and proves it works on a ten second test before you write a single real word. It also does the thing most content courses avoid, which is telling you what this costs per piece, in real numbers, from a pipeline that has actually run.
The most expensive mistake in AI content is not a bad render. It is a perfectly rendered piece that nobody wanted, because the pipeline will happily produce a hundred of those and the cost is the same as producing a hundred good ones. This module puts the deciding before the making. By the end you have a repeatable format rather than a pile of one offs, a bank of ideas you have tested cheaply, and a rule for what to do when one piece outperforms the rest, which is the single moment most people waste.
Everything after this module is automatable. The script is the only part of the pipeline that decides whether any of the rendering was worth doing, and it is the part people rush because it produces nothing they can look at. A script written to be read on a page fails the moment a voice performs it, because the eye forgives what the ear will not. This module writes for the ear, builds attention structurally rather than hoping, and puts a scoring gate in front of production so that a weak script is caught while it still costs nothing.
Two things come out of this module and only one of them is audible. The first is the voiceover, which is what an audience judges within about four seconds. The second is a file of word level timings, which nobody ever hears and which every image cut, every caption and every vertical cut in this pipeline is placed from. Generating audio without that timing data is the single most common way people end up editing by hand for three hours and calling it automation. This module produces both, tunes the voice against what the settings actually do rather than what they are called, and fixes the two faults that make AI narration recognisable.
AI images look fake for a reason that has nothing to do with rendering quality. Models default to the mean: perfect symmetry, ideal light, clean surfaces, centred subjects, uniform beauty, no lens, nothing unplanned in frame. Realism is not extra detail and it is certainly not the word photorealistic. It is disorder that has a cause. This module is a system of nine slots that you fill every time, and a check you run before spending a generation. It is the longest module in the course because it is the one that decides whether your output looks like a business or like somebody with a subscription.
This module puts module five to work on the images a business actually needs. Product shots that a customer believes, lifestyle images set where your customers live rather than somewhere that looks like a stock library, and people who look like the people you sell to. It also covers the two things that separate a usable image set from a folder of nice pictures: generating in batches without burning your quota, and keeping a person or a product recognisably the same across twenty images.
A faceless video is not an animation. It is a sequence of still images cut against a voiceover at a rhythm that reads as movement, and the whole craft is in the rhythm and the cutting rather than in the drawing. This module breaks a script into scenes, locks a style so every image belongs to the same piece, builds a pacing architecture that holds attention without generating hundreds of images, and gets real motion out of stills for nothing. It ends with the thumbnail, because a video nobody clicks is a video nobody watched.
Video has three failure surfaces that a still image does not: there was an operator holding the camera, time passes and things must not change that should not, and there is sound. A generated person is not judged the way a generated photograph is, because a viewer watches a face for several seconds rather than glancing at it, and every one of those seconds is another chance to notice. This module adds four slots on top of the nine, writes UGC that sounds like a person rather than an advert, and deals plainly with disclosure, which is a commercial requirement here and not a matter of conscience.
Carousels and static posts are the least glamorous output in this course and, on the business and service tracks, frequently the most profitable. They cost almost nothing, they are read rather than watched so they survive a silent commute and a weak connection, and they are the format where a written idea can be tested in twenty minutes instead of a day. This module builds them properly: a structure that earns the swipe, a template system so the tenth takes ten minutes, text that stays legible on a phone, and a way to turn one long piece into a month of them.
This is the module where the pipeline becomes a pipeline. Everything so far has produced files in a folder: a script, an audio file, an alignment, a scene map and a set of images. Assembly turns them into a video, and it has to do it the same way every time, from a command rather than from an editor, because a process that involves dragging things on a timeline is a process that produces one video a week no matter how good the rest of your system is. By the end of this module you run one command and a finished file appears.
Everything up to here makes one piece. This module makes a habit, which is a different and harder problem. Nobody fails at this because a render broke. They fail because they made eleven pieces, saw nothing happen, and stopped, which is exactly the point at which the numbers were about to become useful. This module batches production so a week is made in a day, publishes without a person doing it, measures the three numbers that matter on your track and ignores the rest, and deals directly with the thing that actually ends most of these projects.
You now have a machine that produces content. A machine is not a business, and this is the module where the three tracks separate. Read the lesson for your own track properly and skim the other two, because the skill transfers even when the plan does not. The last lesson applies to everybody: what to automate next, which is the decision that determines whether this keeps getting easier or quietly becomes a job you built for yourself.