Where content really starts
In b1 you saw the tool and the two rails. Tonight is the source of everything that follows: the idea. A weak idea drafted beautifully is still a weak piece - polished slop. A strong idea drafted roughly is worth editing into something great. So before you touch a draft, you get two things sharp: the core idea (an insight only you would land) and the content brief (the one page that turns that idea into content only you could make). Everything you build for Cadence from here starts here.
The idea worth scaling 6 min live
Here is the trap: people bring AI a topic - "productivity", "onboarding", "our new feature" - and the model happily returns the average take on it, same as it would for anyone. A topic is a subject. An idea is a subject plus a claim: an insight, a point of view, or a tension you actually believe. AI is brilliant at generating and pressure-testing idea candidates, but the insight at the centre has to be yours. That is the one thing on this whole course the model genuinely cannot supply.
LiveTopic vs idea - the test that saves you from slop3 min▶
The quickest way to tell them apart: a topic can be argued by anyone, an idea has a point of view that someone could disagree with. "Content marketing" is a topic. "Most content teams are drowning because they optimised for volume, not memory" is an idea - it makes a claim, and it could be wrong. That edge is exactly what makes it worth reading. If your line could sit on any brand's blog unchanged, it is a topic, and the model will draft it into perfect, forgettable slop.
- A topic is safe; an idea takes a side. If nobody could disagree, nobody will remember it.
- The insight is yours. AI can list twenty angles on a topic, but it cannot tell you which one you actually believe from experience.
- One real idea beats ten topics. Distinctiveness beats volume - the b1 rail, applied at the source.
The post that finally landed. A creator published "5 tips for better meetings" to silence for months. Then she wrote one line she actually believed - "most meetings are a failure of writing, not talking" - and built a single piece around it. It took a side, so people argued in the comments, quoted it, and shared it. Same topic. The difference was an idea.
Self-studyThree shapes of a strong idea2 min read▶
A strong idea usually takes one of three shapes. Any of them beats a bare topic.
- An insight - something true that is not obvious: "onboarding fails in the first five minutes, not the first week".
- A point of view - a stance you will defend: "we should ship fewer posts and edit them harder".
- A tension - two true things that pull against each other: "AI makes content faster and makes it forgettable".
When you mine ideas later, you are hunting for one of these. If your candidate is none of them, it is still a topic - keep digging.
The content brief 6 min live
An idea in your head is fragile - it drifts into the average the moment you start drafting. The brief is what pins it down. It is a short, plain document that captures the idea, who it is for, the angle, the proof, the format, the voice, and the constraints. Get these seven right and any draft that follows is already halfway to yours. AI can draft the brief in seconds from your notes, but the two rows it cannot fill for you are the insight and the proof - those are yours.
| Brief element | What goes in it |
|---|---|
| Idea | The insight or point of view in one sentence - the claim someone could disagree with. Yours. |
| Audience | Who specifically this is for, and what they already believe or struggle with. |
| Angle | The way in - the hook or frame that makes the idea land for this audience. |
| Proof / specifics | The real examples, numbers, or stories that make it credible and unmistakably yours. Yours. |
| Format | Article, post, thread, newsletter, script - and roughly how long. |
| Voice | How it should sound - two or three adjectives, plus a sample of your writing to match. |
| Constraints | What to avoid, what is confidential, disclosure or rights notes, the deadline. |
LiveWhy the brief is the anti-slop move3 min▶
Slop happens when you hand the model a thin prompt and let it fill the gaps with the average. The brief closes those gaps before drafting starts. When the idea is a real claim, the audience is specific, and the proof is your own, the model has nowhere to be generic - you have already made every decision that matters. The draft becomes execution, not invention. This is the same "the edit is the value" habit from b1, moved one step earlier: decide the substance up front, so editing is polishing, not rescuing.
Self-studyCadence's brief - a worked example2 min read▶
Cadence is the sample product whose content engine this track builds. A filled brief might read:
- Idea: teams adopt Cadence and then quietly go back to spreadsheets - because the problem was never the tool, it was the ritual around it.
- Audience: ops leads who bought a tool that their team stopped using by month three.
- Angle: "the tool was never the problem" - name the uncomfortable pattern out loud.
- Proof: the real drop-off curve from month one to month three; one customer's actual "we went back to Sheets" quote (with permission).
- Format: a 700-word article, plus a short post to tease it.
- Voice: plain, warm, a little contrarian - like a smart colleague, not a brand.
- Constraints: no invented stats; anonymise the customer; disclose nothing confidential.
Notice the idea takes a side and the proof is specific and real. That is a brief the model cannot slop up.
Mining ideas with AI - without the slop 5 min live
Here is the move that keeps AI on the right side of the slop line: never ask it for ideas from nothing. Feed it your raw material - your notes, your half-formed opinions, the thing a customer said last week, the pattern you keep noticing - and ask it to expand that into idea candidates. It becomes a thinking partner that multiplies what is already yours, instead of a slop machine returning the internet's average take. Then you do the one thing it cannot: pick the candidate with a genuine insight.
LiveExpand your material, then you choose3 min▶
The difference between generic-topic slop and real ideas is the input. "Give me content ideas about productivity" gets you the average of everyone. "Here are my messy notes from three client calls - find the tensions and non-obvious patterns, and turn each into an idea that takes a point of view" gets you candidates rooted in something real. The model widens; you decide. Reject anything that could have been written about any company - keep only what carries your fingerprint.
- Bring raw material, not a blank topic. Notes, transcripts, opinions, a screenshot of a customer message - the more specific, the less generic the output.
- Ask for tensions and non-obvious patterns, not "10 ideas". You want the shapes from Part 1: insight, point of view, tension.
- You pick, and you say why. The candidate you choose should be one you could defend from experience. If none qualify, feed it more of your material.
Mine your ideas, pick the one with a POV ★ 13 min · everyone builds
Turn your own experience into idea candidates with AI, then do the human part: pick the one with a genuine insight and sharpen it to a single sentence. This sentence becomes the top line of your Cadence brief.
Brain-dump 5-10 messy lines from your own world - things you have noticed, opinions you hold, a customer quote, a pattern that annoys you. Do not tidy them.
Paste them into ChatGPT or Claude with the mining prompt below. Ask for candidate ideas that take a point of view, plus which are the riskiest.
Read the candidates critically. Reject every one that could appear on a competitor's blog unchanged. Circle the one you could defend from experience.
Sharpen it to one sentence with a claim someone could disagree with. Cut hedging ("can", "might", "some"). Make it a line you would put your name on.
Run the slop test: could only you have written this? If yes, you have your core idea. Keep it open - Demo 2 wraps a brief around it.
Write Cadence's content brief ★ 8 min · build your own
Wrap your sharpened idea in a full brief for Cadence - idea, audience, angle, proof, format, voice, constraints. AI drafts the scaffolding from your idea; you fill and own the insight and the proof. Keep this file: b3 feeds it straight into the repurposing simulator.
Give the AI your one-sentence idea from Demo 1, plus a couple of lines about Cadence and who you are writing for.
Ask it to draft the seven-part brief. Let it propose audience, angle, format, and voice - it is good at these from your idea.
Now take the pen. Replace the idea row with your exact sentence, and fill the proof row with real specifics only you have - a number, a quote, a story. No invented facts.
Tighten the constraints: what to avoid, what is confidential, any disclosure or rights notes, the deadline. This is where the b1 rails live in the brief.
Save it as your Cadence brief. You will hand this same brief to the b3 simulator and build long-form, social, and visual from it in later sessions.
The brief that outlived the draft. A content lead started writing every piece from a one-page brief instead of a blank page. The first draft got faster, but the real win came later: when she repurposed the piece into a thread, a newsletter, and a talk, the brief kept all of them on the same idea and proof. One sharp brief powered a month of aligned content. That is exactly what b3 does with yours.
Try it yourself - this week ◐ 30-45 min total
- Finish both demos if you did not complete them live - especially the sharpened one-sentence idea. That line is the seed of everything in b3-b10.
- Run the mining prompt on a second batch of your own notes. Keep the two idea candidates you would actually defend.
- Finalise your Cadence content brief - all seven rows, with real proof in the proof row. Save the file; b3 needs it.
- Take one old topic you have written before and rewrite it as an idea that takes a side. Notice how much sharper the brief becomes.
- Optional: run the slop test on three ideas in your backlog. Cut anything a competitor could publish unchanged.
Official sources covered
Taught from the content-craft canon and 2026 AI content practice. Certificates and paid tooling stay official. This session covers:
Three questions before you go 🎯 ◐ 90 seconds
1 · What separates an idea from a topic?
A topic is safe and could sit on any brand's blog. An idea takes a side, so it is memorable - and only you can supply the insight at its centre.
2 · What does a content brief contain?
The brief pins the idea down before drafting so the model has nowhere to be generic. Get the seven elements right and the draft is already halfway yours.
3 · When mining ideas with AI, what stays human?
Feed AI your own notes and let it expand them into candidates, but the insight you commit to and the real proof are yours. Outsource those and you have briefed slop.