How this track works
Ten sessions, all hands-on, all building one thing: Cadence's content engine - the ongoing content operation for a sample product you may have branded and marketed in the sibling courses. Tonight you learn what the tool is; then b2 sharpens the idea and brief, b3 turns one idea into many formats with a live simulator, and b4-b10 build the long-form, social, visual, video, pipeline, and governance of a real content operation. No code, ever. Two rails run through all ten, starting now.
What AI actually is - for a content creator 6 min live
A large language model predicts the most likely next words. For content that makes it a phenomenal draft-and-variation engine - fast, tireless, multi-format - and a dangerous one, because "most likely" is another word for average. Left alone it produces content that is fluent, confident, and instantly forgettable. The speed is real; so is the slop. The whole craft is turning one into content and not the other.
LiveThe slop problem - why fluent is not the same as good3 min▶
Because the model generates the most likely phrasing, everyone prompting it lazily gets the same beige output: "In today's fast-paced world...", the tidy three-bullet list, the confident summary that says nothing. It reads fine and lands nowhere. Multiply that by volume and you have flooded your own channels with content that trains your audience to scroll past you. Slop is not a typo problem; it is an averageness problem.
- Fluent is cheap now. Grammatical, structured, confident text is free. That means it is no longer a signal of quality - or of you.
- It invents facts and quotes. Confident statistics and testimonials that do not exist. In content, publishing one is a credibility hit you do not get back.
- The edit is the value. AI gets you to a fast, generic 70%; the specific detail, the real story, the point of view, and the truth-check - the 30% that earns the reader - is human.
The brand that published its way into silence. A team used AI to triple its output overnight. Six months later engagement had cratered - not because the content was wrong, but because it was indistinguishable from everyone else's AI content. They cut volume by half, edited hard, and added a real point of view. Reach recovered. Speed had been the easy part; the edit was the job.
Self-studyThe can / cannot table - pin this to your monitor2 min read▶
| Trust it to... | Never trust it to (the human does) |
|---|---|
| Draft fast and generate many variations | Produce facts, stats, or quotes from memory |
| Repurpose one idea across formats | Have a real point of view |
| Outline and structure a piece | Sound like you by default |
| Adapt tone and length on request | Judge what is actually distinctive |
| Suggest hooks and angles to pick from | Decide what is true and what ships |
The left column is a build-along later in this track. The right column is why every session pairs AI's speed with your edit, your voice, and your judgment.
The content AI tool landscape, 2026 5 min live
The big 2026 shift is multimodal: text, image, video, and audio generated together from a single brief. The tools cluster into three layers - and the right question is which layer fits the format, and whether it is safe for your inputs.
LiveText, specialized, and multimodal3 min▶
- General assistants (ChatGPT, Claude, Gemini) - your workhorses for ideas, drafts, repurposing, and structure. Where most of this track lives.
- Specialized by format - image (Midjourney, DALL-E), video (Runway and similar), voice (ElevenLabs). You will meet these in b6 and b7.
- Multimodal from one brief - the 2026 shift: one brief produces text, visuals, a video script, and audio together, compressing multi-channel production by around 60-70%. Powerful, and exactly where the anti-slop edit matters most.
Self-studyThe 2026 landscape at a glance2 min read▶
- Text: ChatGPT, Claude, Gemini for ideation, drafting, repurposing, and editing support.
- Image: Midjourney, DALL-E, and design suites for graphics, covers, and social visuals.
- Video + audio: Runway and similar for video, ElevenLabs for voice - the fastest-moving, most rights-sensitive layer.
- The trend: multimodal production from a single brief, refined with brand controls and real performance feedback. The bottleneck is no longer making content - it is keeping it distinctive and true.
The two rails: own the voice, respect the rights 4 min live
Everything else in this course is technique. These two rails are non-negotiable, and they are what separate a creator who uses AI to make more of what only they could make from one who floods the world with forgettable slop.
LiveRail 1 - own your voice and your claims2 min▶
AI drafts; you own every word, fact, and point of view that ships under your name. That means the human edit is non-negotiable: no fabricated statistics or quotes, real specifics over generic filler, and a genuine perspective the model cannot supply. Distinctiveness beats volume every time - one piece that sounds like you outperforms ten that sound like everyone.
The edit is the author. Anyone can generate a draft now; that is not the value you add. The value is the sentence only you would write, the story only you lived, the number you actually checked. Treat the AI draft as raw clay and the edit as the sculpture. Ship the sculpture, never the clay.
LiveRail 2 - respect the rights (IP, likeness, disclosure)2 min▶
AI-generated text, images, video, and voice carry real exposure: copyright and training-data questions, likeness and right-of-publicity for real people, and disclosure rules for AI content. Use rights-clear inputs, check the tool's commercial-use terms, disclose AI where required, and adopt provenance practices as they mature.
- Inputs must be clear. Do not feed others' copyrighted work or a real person's likeness without rights.
- Outputs need a check. AI images can echo a living artist's style or a recognizable face - review before publishing, and mind the tool's terms.
- Disclose + provenance. Where AI content or endorsements require disclosure (FTC), do it; content-credential and authenticity practices are becoming table stakes.
Slop vs edited - see the edit earn the reader ★ 12 min · everyone builds
The single most important habit in the course, in one experiment. Take a raw AI draft, then run a real edit pass, and watch generic slop become something worth publishing.
Ask any chat AI for a short piece on a topic you know - a post, a paragraph, an intro. Take the raw draft as-is.
Read it critically: circle every generic phrase ("in today's world", "game-changer"), every claim you cannot verify, and every place it could be anyone.
Now edit: cut the filler, replace one vague claim with a real specific from your own experience, add one sentence only you would write, and verify any fact.
Compare the two. The edited version is shorter, truer, and unmistakably yours - and it took minutes. That delta is the whole job.
Name the three edit moves you made. Those become your default anti-slop pass, which you will formalize in b9.
Write your content AI ground rules ★ 10 min · build your own
A reusable artifact you will keep: a short "how we make content with AI" card. You draft it with AI, then edit it - which is rail one in miniature.
Give the AI the two rails and your context: what you publish, the channels, and what is confidential or client-owned.
Ask for a one-page card with three lists: Green light (safe uses - drafts, variations, repurposing, structure), Red light (never: publishing unedited, fabricated facts/quotes, others' IP or likeness without rights), and Always (human edit, verify facts, keep the voice, disclose + check rights).
Edit it hard. Add your real channels and the lines your brand or clients draw. The AI draft is 70%; your judgment is the 30% that matters.
Share it with your team. You will refine it in the leader session a3 (voice + anti-slop) and the creator finale b10 (governance + publishing).
The card that saved a team from its own speed. A content team adopted AI and quietly started shipping unedited drafts to hit volume targets. A one-page card - "nothing publishes without a human edit and a rights check" - reset the standard. Output stayed high; quality came back. The rule was cheaper than the reputation it protected.
Try it yourself - this week ◐ 30-45 min total
- Finish both demos if you did not complete them live - especially slop-vs-edited. Feeling the edit earn the reader is the lesson.
- Run the edit-pass prompt on one real AI draft this week. Keep the three edit moves you made most often.
- Finalise your one-page content AI ground rules and share it with a teammate.
- Screenshot the five near-identical posts from the Part 1 "feel the slop" prompt - that is your reminder of what volume-without-edit looks like.
- Optional: pick one channel where your content has drifted toward the average, and note what a real point of view would add.
Official sources covered
Taught from public AI content curricula and content-craft practice. Certificates and paid tooling stay official. This session covers:
Three questions before you go 🎯 ◐ 90 seconds
1 · What is "AI slop"?
The model drafts at the most-likely average, so lazy prompting yields interchangeable, forgettable content. Fluent is now cheap; distinctiveness is the signal.
2 · Where is the value a creator adds now that AI drafts are free?
AI gets you a generic 70%; the edit is the 30% that earns the reader. Ship the sculpture, not the clay.
3 · Before publishing AI-generated visuals or video, you must...
Rail two. AI media carries copyright, likeness, and disclosure exposure. Clear the inputs, review the outputs, check the tool's terms, and disclose where required.