How this track works
Ten sessions, all hands-on, all building one thing: the whole brand foundation for Cadence, a sample AI note-taker - the prequel to the campaign you could run in the AI + Marketing course. Tonight you learn what the tool is; then b2 sets positioning and strategy, b3-b7 build personality, naming, voice, visual direction, guidelines, and messaging, and b8-b10 activate, monitor, and govern it. No code, ever. Two rails run through all ten, starting now.
What an LLM actually is - for a brand builder 6 min live
A large language model predicts the most likely next words from a huge slice of the internet. For brand work that cuts two ways: it is a tireless engine for exploration, drafts, and pattern - and it pulls everything toward the average, which is the exact opposite of what a brand is for. A brand exists to be different; AI, left alone, makes you the same as everyone.
LiveThe sameness trap - the one brand risk unique to AI3 min▶
Every other AI risk (fabrication, IP) shows up in marketing too. This one is brand's alone: because the model predicts the most likely phrasing, everyone who prompts it the same lazy way gets the same average answer. Ask ten founders for "a bold, innovative tagline" and you get ten variations of the same beige sentence. The result is category-wide homogenization - and a brand that blends in has failed at its only job.
- The model pulls to the mean. Distinctive is, by definition, unlikely - so you have to steer hard toward it.
- A documented system is the steering. A locked positioning, a real personality, and voice rules with a "never say" list turn the average engine into a Cadence-shaped one.
- Without a system, AI is a dilution machine. With one, it is a scaling machine. Same tool, opposite outcome - which is the whole reason this course exists.
The category that started sounding like one company. When a wave of startups all reached for the same AI tools with the same thin prompts, their sites converged - same gradients, same "empower your team to do their best work", same everything. The brands that stood out were the ones with a documented system the AI had to obey. Distinctiveness did not come from the model. It came from what they fed it.
Self-studyThe can / cannot table - pin this to your monitor2 min read▶
| Trust it to... | Never trust it to (keep human) |
|---|---|
| Generate naming, tagline, and angle options to react to | Decide what the brand stands for |
| Draft on-brand copy once the system exists | Judge which option is actually distinctive |
| Explore visual moods and directions | Own the brand's meaning and story |
| Audit assets for consistency at scale | Produce facts, stats, or legal/IP clearance |
| Summarize competitor and audience research | Give final sign-off on anything public |
The left column is a build-along later in this track. The right column is why every session pairs AI's speed with your judgement - the brand soul never gets outsourced.
The brand AI tool landscape, 2026 5 min live
The right first question is not "which tool" but "is this safe for my brand and my data, and does it stay on-system?" Three layers, and the line between them is about control and IP, not raw quality.
LiveGeneral assistants, brand GPTs, and image tools3 min▶
Each layer has a job in brand work:
- General assistants (ChatGPT, Claude, Perplexity) - competitor and audience research, naming and voice exploration, consistency audits.
- Brand-native + brand GPTs - the big unlock: a custom GPT or tool loaded with your positioning, personality, and voice rules generates on-system every time. You build Cadence's in session b6.
- Image + direction tools (Midjourney, DALL-E, Canva) - moodboards and visual direction to align a team and brief a designer. Direction, not final production art.
Self-studyThe 2026 landscape at a glance2 min read▶
- Research + strategy: ChatGPT, Claude, Perplexity for competitor scans, audience sentiment, and category mapping.
- Voice + content at scale: brand GPTs and tools like Jasper that hold a documented voice so output stays on-brand.
- Visual direction: Midjourney, DALL-E, Canva for moodboards and exploration - with IP and likeness caution (session b5, b10).
- The 2026 shift: documented brand guidelines - especially verbal, with a "never say" list - are now a requirement, because they are what lets AI scale a brand without diluting it.
The two rails: own the soul, protect the rights 4 min live
Everything else in this course is technique. These two rails are non-negotiable, and they are what separate a brand that uses AI to get sharper from one that dissolves into the category average.
LiveRail 1 - humans own the brand soul and the claim2 min▶
AI explores, drafts, and audits; humans decide what the brand means, judge what is truly distinctive, and give final sign-off. The strategy, the story, and the taste stay yours - and the antidote to AI sameness is a documented system that forces the model off the average and onto your brand.
The rebrand that outsourced the soul. A team let AI generate its entire positioning from a one-line prompt and shipped it. It was competent, fluent, and indistinguishable from three competitors who had done the same. They had automated the one part - meaning - that was supposed to be theirs. Rail one exists so the machine sharpens your judgement instead of replacing it.
LiveRail 2 - protect the rights (IP, likeness, data)2 min▶
AI-generated names, logos, and images carry real legal exposure: trademark conflicts, copyright questions over training data, and right-of-publicity or likeness issues with real people. Clear anything before use, keep a human review, and mind the data you feed in.
- Names and logos need clearance. An AI name may already be someone's trademark - always check before you commit.
- Images carry rights risk. Style-mimicry of a living artist, or a recognizable face, can be a legal problem. Check the tool's commercial-use terms.
- Disclosure where required. FTC/ASA rules on AI and endorsements apply to brand content too.
Sameness vs distinct - see the average, then beat it ★ 12 min · everyone builds
The single most important habit in the course, in one experiment. Ask AI for brand language with a lazy prompt, then with a real brief, and watch generic average copy become something only Cadence could say.
Open any chat AI. Pick a safe subject - our sample product Cadence, or one you admire.
Average ask: "give me a bold tagline and positioning" with no brief. Note how it could belong to any product in the category.
Systematized ask: paste a short brief - the specific audience, the real point of difference, a personality word, and one thing the brand would never say - and ask again.
Compare. The second is sharper, stranger, more ownable - because you steered the model off the average. That delta is your brand system doing its job before it even exists on paper.
Now push it: add a "never say" rule (e.g. no "revolutionary", no "seamless") and re-run. Watch the beige words disappear. That is a voice guideline working, and you will build a real one in b4.
Write your brand's AI ground rules ★ 10 min · build your own
A reusable artifact you will keep: a short "how we use AI on the brand" card. You draft it with AI, then edit it - which is rail one in miniature.
Give the AI the two rails and your context: brand, the tools you have (or have not) approved, and what is confidential (unreleased identity work, brand assets).
Ask for a one-page card with three lists: Green light (safe uses - research, exploration, drafting to the system), Red light (never: outsourcing the strategy, shipping AI names/logos without clearance), and Always (steer off the average with the system, human sign-off, check IP + disclosure).
Edit it hard. Add your real tools and who signs off. The AI draft is 70%; your judgement is the 30% that matters.
Keep it. You will refine it in the leader session a3 (integrity + IP) and the builder finale b10 (governance).
The card that kept a brand a brand. A team adopted AI fast and their output quietly drifted toward the category average within a quarter. A one-page rule - "every AI asset must be steered by our system and pass a human distinctiveness check" - pulled it back. Speed stayed; sameness left. That is the entire promise of this course in one habit.
Try it yourself - this week ◐ 30-45 min total
- Finish both demos if you did not complete them live - especially sameness-vs-distinct. Feeling the average yourself beats being told about it.
- Run the average prompt on your own brand or category. Screenshot the beige result - that is your reminder of what everyone without a system is shipping.
- Finalise your one-page brand AI ground rules and share it with a colleague.
- Jot the one thing your brand should NEVER say. You will use it in b4 to build a real voice guideline.
- Optional: skim your current brand guidelines (or note that you need some). Bring one gap to session b2.
Official sources covered
Taught from public brand-AI curricula, brand-strategy canon, and vendor documentation. Certificates and paid tooling stay official. This session covers:
Three questions before you go 🎯 ◐ 90 seconds
1 · What is the brand risk unique to AI that this course is built around?
The model predicts the most likely phrasing, so lazy prompting yields the category average. A brand exists to be different - so distinctiveness must be engineered with a system.
2 · What actually turns a generic AI assistant into an on-brand one?
The system is the steering. Without it AI dilutes; with it, AI scales the brand. Same tool, opposite outcome.
3 · Which part of brand work must stay human?
Rail one: AI explores and drafts; humans own the soul, the taste, and the claim. Outsource the meaning and you become the average.