Make AI Look and Sound Like Your Business — Make AI Look and Sound Like Your Business

Session 02 · Extended talk track · Slide 1

Make AI Look and Sound Like Your Business

Expanded talk track

Last time, we focused on asking clearly, checking the work, and keeping authority with the right person. Today we apply that operating model to marketing. The question is not simply whether AI can produce more copy or images. The question is whether the result still looks and sounds like your business. We will make something together early. The example is VoltGuard Electric, a fictional training brand, so nobody’s real business is being put on the spot. We will start with a thin request, inspect the generic result, then add the audience, promise, voice, visual rules, factual boundaries, and next step. The goal is not a perfect campaign. It is a repeatable method for giving AI enough real business context that a person can judge the draft against a standard.

Land this

AI can help make more marketing. Your brand makes it yours.

Participant recap, exercise, and public sources

Recap

AI can help you make more marketing. Your brand makes it yours.

Try it

Write one sentence describing what customers should recognize as unmistakably yours.

Resources

Use the included NCCC Session 2 Prompt Pack for the reusable prompts and review checklist.

Public sources

- No external source is required for this slide. The examples and exercises are contained in the deck and prompt pack.

1. Make AI Look and Sound Like Your Business

Make AI Look and Sound Like Your Business — Make AI Look and Sound Like Your Business

Expanded talk track

Last time, we focused on asking clearly, checking the work, and keeping authority with the right person. Today we apply that operating model to marketing. The question is not simply whether AI can produce more copy or images. The question is whether the result still looks and sounds like your business. We will make something together early. The example is VoltGuard Electric, a fictional training brand, so nobody’s real business is being put on the spot. We will start with a thin request, inspect the generic result, then add the audience, promise, voice, visual rules, factual boundaries, and next step. The goal is not a perfect campaign. It is a repeatable method for giving AI enough real business context that a person can judge the draft against a standard.

Land this

AI can help make more marketing. Your brand makes it yours.

Participant recap, exercise, and public sources

Recap

AI can help you make more marketing. Your brand makes it yours.

Try it

Write one sentence describing what customers should recognize as unmistakably yours.

Resources

Use the included NCCC Session 2 Prompt Pack for the reusable prompts and review checklist.

Public sources

- No external source is required for this slide. The examples and exercises are contained in the deck and prompt pack.

2. I will call out what not to do.

Make AI Look and Sound Like Your Business — I will call out what not to do.

Expanded talk track

I am going to point at common AI-generated styles and explain why they can fail. If you have used one of them, that does not make you foolish or bad at marketing. The tools make these defaults easy to reach, especially when the request contains little about the actual business. Recognition is useful evidence. If you immediately see your own recent post in an example, the lesson is not that you should feel embarrassed. It tells us the workflow needs more of your audience, voice, visual system, and factual boundaries inside it. We can critique an output without judging the person who made it. We also need to avoid the reverse mistake: assuming every brush font, polished photo, or bright gradient proves AI involvement. The problem is not a single style choice. The problem is a result that could belong to almost anyone.

Land this

Critique the output. Use the critique to strengthen the system.

Participant recap, exercise, and public sources

Recap

I can critique the output without judging the person who made it.

Try it

List one common AI style you want to avoid. Describe the business reason, not just the aesthetic dislike.

Resources

Use the included NCCC Session 2 Prompt Pack for the reusable prompts and review checklist.

Public sources

- No external source is required for this slide. The examples and exercises are contained in the deck and prompt pack.

3. Different businesses. Same everything.

Make AI Look and Sound Like Your Business — Different businesses. Same everything.

Expanded talk track

Here are three businesses with completely different reasons to buy: a food truck, a dentist, and a hair salon. Yet the ads share the same brush-style headline, vague promise, familiar layout, and empty call to “discover more.” Cover the business names and ask whether anything else tells you who made each piece. Each design choice might be acceptable by itself. The problem appears when the same visual costume and generic language show up across unrelated businesses. Without specific direction, a model falls back on familiar patterns. General patterns are not the same as your practices, customers, or promise. This is why adding only a business name rarely solves the problem. The system needs the differences that actually matter: who the message is for, what they need, how the business speaks, and what action makes sense next.

Land this

The real problem is not “made with AI.” It is “could belong to anybody.”

Participant recap, exercise, and public sources

Recap

The problem isn’t just “made with AI.” It’s “could belong to anybody.”

Try it

Cover the names in three recent posts. Note which details still reveal the business behind each post.

Resources

Use the included NCCC Session 2 Prompt Pack for the reusable prompts and review checklist.

Public sources

- No external source is required for this slide. The examples and exercises are contained in the deck and prompt pack.

4. These are the tells.

Make AI Look and Sound Like Your Business — These are the tells.

Expanded talk track

The examples on this slide are newly generated fictional teaching recreations, not screenshots of real campaigns or participant work. Look for the recurring visual grammar: oversized brush lettering, glossy lighting, implausibly perfect people, decorative filler text, and a vague aspirational promise. People often say they can immediately recognize an AI ad. Usually they are recognizing a cluster of defaults before they recognize the business. The small copy may even melt into the image because an image generator treats letters as visual texture rather than editable language. Do not turn this into a rule that polished photography or brush lettering always means AI. The useful question is whether the tool’s habits are louder than the brand. If the audience notices the production method before the offer, the creative is doing the wrong job.

Land this

When the tool becomes the main character, the business disappears.

Participant recap, exercise, and public sources

Recap

When the tool’s habits are louder than the brand, the tool becomes the main character.

Try it

Audit one recent image for generic headlines, polished stock-like people, filler text, and vague promises.

Resources

Use the included NCCC Session 2 Prompt Pack for the reusable prompts and review checklist.

Public sources

- https://www.kantar.com/Inspiration/Advertising-Media/Rethinking-AI-generated-advertising

5. Demo 1: start without brand rules.

Expanded talk track

We are starting the live demo now. VoltGuard Electric is fictional, and the result is for training only. The first request asks for a social post about thermal imaging scans. It includes a headline, body, and call to action. It also tells the tool not to invent prices, credentials, reviews, statistics, response times, savings, safety records, or guarantees. That factual boundary is important, but the request still says almost nothing about the customer, the company’s voice, or the visual system. Generate once and read only the headline and one sentence. Then mark every line that could belong to almost any commercial electrician. If the output happens to be good, acknowledge it. The test is still whether it fits this specific buyer and brand, not whether the prose sounds polished.

Land this

A business name and service category are not enough direction.

Participant recap, exercise, and public sources

Recap

A business name is not enough direction.

Try it

Use Prompt 1 in the included prompt pack. Mark every line that could belong to a competitor.

Resources

Use the included NCCC Session 2 Prompt Pack for the reusable prompts and review checklist.

Public sources

- No external source is required for this slide. The examples and exercises are contained in the deck and prompt pack.

6. Now give it a business to sound like.

Make AI Look and Sound Like Your Business — Now give it a business to sound like.

Expanded talk track

Now we add the decisions that make VoltGuard recognizable. The audience is facility managers, restaurant operators, and manufacturers who need electrical work planned around an operating business. The service is a thermal imaging scan that helps an electrician inspect equipment for heat patterns that may require attention. It does not guarantee prevention or prove that equipment is safe. The voice is direct, confident, and calm under pressure. The visual system uses dark structure, a yellow accent, approved typography, and the unchanged logo. The factual boundary rejects response-time promises, savings percentages, safety records, licenses, prices, and testimonials unless evidence has been supplied. The next step is a facility assessment, not a miracle claim. This is the core brief: audience, promise, voice, visual rules, factual boundaries, and next step.

Land this

Useful specificity comes from business decisions, not magic prompt words.

Participant recap, exercise, and public sources

Recap

Audience + promise + voice + visual rules + factual boundaries + next step.

Try it

Draft six brand rules: audience, promise, voice, visual system, factual boundaries, and next step.

Resources

Use the included NCCC Session 2 Prompt Pack for the reusable prompts and review checklist.

Public sources

- No external source is required for this slide. The examples and exercises are contained in the deck and prompt pack.

7. Same tool. Different direction.

Make AI Look and Sound Like Your Business — Same tool. Different direction.

Expanded talk track

Run the branded request in the same conversation, or use the prepared teaching examples if the live tool is unavailable. Compare the two versions. What changed about the customer, the service, the language, the next step, and the claims the draft avoided? The improved example should speak to an operating business. It should use concrete language, respect the calm tone, and avoid claiming that a scan prevents every outage or proves equipment is safe. It remains a draft. Better context gives us something we can judge against an explicit standard. It does not eliminate human review. Be transparent about provenance. The prepared examples on the slide are authored teaching samples. They should not be described as the live model’s output. If you run the demo, identify today’s actual output separately.

Land this

Brand rules turn “make it better” into choices we can inspect.

Participant recap, exercise, and public sources

Recap

Better context gives us something we can judge against our standards.

Try it

Run Prompt 2 from the prompt pack. Compare the result with the thin-prompt draft and keep only supported improvements.

Resources

Use the included NCCC Session 2 Prompt Pack for the reusable prompts and review checklist.

Public sources

- No external source is required for this slide. The examples and exercises are contained in the deck and prompt pack.

8. Marketing is the whole loop.

Make AI Look and Sound Like Your Business — Marketing is the whole loop.

Expanded talk track

A post is one part of a larger system. The offer says what we are asking someone to choose. Creative makes the offer understandable. Distribution reaches the intended audience. The destination explains the next step. Proof builds confidence. Follow-up continues the conversation. Measurement tells us what to improve. The center should remain stable: who we are, who we serve, and what we promise. If the ad promises a calm assessment but the landing page uses fear and pressure, the system breaks. If the post attracts attention but the next step is confusing, better copy cannot repair the journey. This loop is a teaching synthesis, not a quotation from a single source or a claim that every business must use the same campaign model. Its purpose is to stop us from treating one generated post as the entire marketing strategy.

Land this

A good post cannot fix a broken next step.

Participant recap, exercise, and public sources

Recap

A good post cannot fix a broken next step.

Try it

Map one real campaign across offer, creative, distribution, destination, proof, follow-up, and measurement.

Resources

Use the included NCCC Session 2 Prompt Pack for the reusable prompts and review checklist.

Public sources

- No external source is required for this slide. The examples and exercises are contained in the deck and prompt pack.

9. Not everyone. Someone.

Make AI Look and Sound Like Your Business — Not everyone. Someone.

Expanded talk track

“Everyone” is usually too broad to guide a useful message. Specificity does not mean excluding every other customer. It means describing one person or business clearly enough that they can recognize the message as relevant. Name the customer and the job they are trying to complete when they encounter the message. A facility manager trying to schedule an inspection around production needs a different emphasis from a homeowner comparing emergency service options. The service may overlap, but the concern, evidence, and next step change. This slide uses wording that can be traced to the cited public source rather than relying on a remembered paraphrase. That is a small example of the verification habits from Session 1: if we repeat someone else’s idea, we should be able to show where it came from.

Land this

Specific means the right person can recognize that the message is for them.

Participant recap, exercise, and public sources

Recap

Specific does not mean small-minded. It means someone can recognize that it is for them.

Try it

Name one specific customer and the job they are trying to complete when they encounter your message.

Resources

Use the included NCCC Session 2 Prompt Pack for the reusable prompts and review checklist.

Public sources

- https://seths.blog/2006/04/ode_how_to_tell/

10. Same Mark. Different promise.

Make AI Look and Sound Like Your Business — Same Mark. Different promise.

Expanded talk track

One person can operate more than one brand without making those brands interchangeable. The personal training identity is designed for education: approachable, clear, and easy to enter. StarForge covers a broader technical promise around software, automation, implementation, and guidance, so its visual system can feel more assertive and technical. There may be customer overlap, but the promise and entry point differ. That difference should affect the language, image direction, call to action, and level of detail. Swapping only the logo would not preserve the meaning of either system. Personal taste can be a valid starting point. Colors, type, and visual references often begin with what the owner genuinely likes. The professional work is turning those preferences into consistent rules other people can apply.

Land this

The brand should fit the promise, not merely the person selling it.

Participant recap, exercise, and public sources

Recap

The brand should fit the promise—not just the person selling it.

Try it

If you have more than one offer or brand, write a separate promise and entry point for each.

Resources

Use the included NCCC Session 2 Prompt Pack for the reusable prompts and review checklist.

Public sources

- No external source is required for this slide. The examples and exercises are contained in the deck and prompt pack.

11. Vector Wood-cut Neo Industrial

Make AI Look and Sound Like Your Business — Vector Wood-cut Neo Industrial

Expanded talk track

There was no single established style name that captured what I wanted. I described the visual references I liked, supplied examples and search terms, and reacted to repeated iterations. Research helped identify ingredients such as Silver Age comic inking, technical and educational graphics, poster composition, geometric structure, and neon color. The important part is not the unusual name. It is the process of turning taste into language another person or tool can use. You can recognize that an image feels right before you know the design vocabulary. Collect examples, explain the specific feature you like, and record what changes your reaction. The word “vector” in this style name describes the aesthetic. It does not certify that every displayed image is a production-ready vector file. References to poster history describe composition, not political endorsement.

Land this

You can know what you like before you know the vocabulary.

Participant recap, exercise, and public sources

Recap

You can know what you like before you know the vocabulary.

Try it

Collect three visual references you genuinely like. Explain the specific composition, texture, or color choice in each.

Resources

Use the included NCCC Session 2 Prompt Pack for the reusable prompts and review checklist.

Public sources

- No external source is required for this slide. The examples and exercises are contained in the deck and prompt pack.

12. Tracking: a VCR or a font?

Make AI Look and Sound Like Your Business — Tracking: a VCR or a font?

Expanded talk track

“Plus-50 tracking? What is this, a VCR or a font?” I did not begin with the terminology. I knew I wanted more breathing room between the letters. Learning the setting gave me a way to reproduce the feeling. Tracking adjusts spacing across a run of text. Kerning adjusts the space between particular letter pairs. A small change can make a familiar typeface feel friendly, formal, technical, or urgent. It can also damage readability when pushed too far. The practical exercise is simple. Choose one real headline and test two spacing settings. Look at it at the actual size where customers will see it. Keep the version that supports the intended tone and remains easy to read. Record the chosen setting so the next piece does not begin from zero.

Land this

A tiny setting can become a repeatable brand decision.

Participant recap, exercise, and public sources

Recap

A tiny setting can make a familiar font feel specific to a brand.

Try it

Choose one headline and test two tracking settings. Keep the version that supports the intended tone and stays readable.

Resources

Use the included NCCC Session 2 Prompt Pack for the reusable prompts and review checklist.

Public sources

- No external source is required for this slide. The examples and exercises are contained in the deck and prompt pack.

13. Map the voice. Don’t ask for “better.”

Make AI Look and Sound Like Your Business — Map the voice. Don’t ask for “better.”

Expanded talk track

“Make it better” gives the tool no standard. Instead, provide examples and explain your reactions. Ask the AI to identify patterns in what you like and dislike, then test those patterns with contrasting samples. A useful voice trait changes the writing. “Reassuring” might mean calm language, concrete next steps, and no fake certainty. “Expert” might mean precise terms with a plain-language explanation, not jargon used to impress. “Direct” might mean the main action appears in the first sentence and the draft removes empty introductions. Treat your corrections as evidence. Save approved examples, phrases to avoid, sentence habits, and factual boundaries. Reuse the rules instead of asking the tool to rediscover your voice every time.

Land this

Your corrections are brand data. Capture them.

Participant recap, exercise, and public sources

Recap

Your corrections are brand data. Capture them.

Try it

Use Prompt 3 in the prompt pack. Define each voice trait as an observable writing behavior.

Resources

Use the included NCCC Session 2 Prompt Pack for the reusable prompts and review checklist.

Public sources

- No external source is required for this slide. The examples and exercises are contained in the deck and prompt pack.

14. A brand kit shows the decisions.

Expanded talk track

A useful brand kit does more than collect attractive assets. It answers what to use, where to use it, and how much. Include the approved logo variants, color values, font roles, spacing expectations, and examples that show the system in practice. Type should have jobs. One style carries the main message, another explains it, and a technical label may serve a narrower role. Color also needs hierarchy. A strong accent can direct attention without becoming paragraph text or covering every surface. The StarForge examples on this slide are teaching material showing a horizontal logo, stacked logo, mark, type roles, and color hierarchy. The comparison asset is labeled as a draft reconstruction. That label matters because a clean-looking training comparison is not automatically an approved production master.

Land this

A kit is useful when another person can make a consistent choice with it.

Participant recap, exercise, and public sources

Recap

A usable kit answers what to use, where, and how much.

Try it

Create a one-page kit with approved logo files, color values, font roles, spacing rules, and examples.

Resources

Use the included NCCC Session 2 Prompt Pack for the reusable prompts and review checklist.

Public sources

- No external source is required for this slide. The examples and exercises are contained in the deck and prompt pack.

15. Looks like a logo ≠ logo-ready.

Expanded talk track

An image can have an SVG filename and still contain only an embedded raster picture. At ordinary size it may look completely fine. The limitation appears when a printer needs separable shapes, when the mark is enlarged, or when someone must change one element without rebuilding the image. A true vector-path reconstruction stays sharp because the shapes are mathematically defined. That does not make every reconstruction an approved master. Type, spacing, colors, geometry, and export settings still need visual review against the source. Image generation is excellent for direction, exploration, and mockups. A production logo needs a clean master and the correct set of exports. Raster artwork can also print well at sufficient resolution, so the lesson is not that raster is always bad. The issue is whether the file supports the job it is expected to do.

Land this

Inspect the asset, not only the filename or screen preview.

Participant recap, exercise, and public sources

Recap

Inspect the file—not just the filename or how it looks on screen.

Try it

Inspect your logo master at high zoom. Confirm whether it contains editable vector paths or only an embedded raster image.

Resources

Use the included NCCC Session 2 Prompt Pack for the reusable prompts and review checklist.

Public sources

- No external source is required for this slide. The examples and exercises are contained in the deck and prompt pack.

16. Give AI rules it can actually use.

Make AI Look and Sound Like Your Business — Give AI rules it can actually use.

Expanded talk track

A brand kit is stronger than a descriptive prompt when the working tool can keep the approved logo, editable text, fonts, colors, layout, and safe zones as separate elements. The goal is not to regenerate a logo or bake the headline into an image. The goal is an editable draft inside a system that preserves the brand decisions. AI integrations and connectors change over time and vary by account. Do not treat a documented integration as proof that a specific connected workflow will work in every account today. A reliable instruction asks the tool to show which kit and assets it can actually access, list substitutions, and stop before publishing or spending. The workflow still holds when a connector is unavailable: start from an approved template, place approved copy and assets into editable layers, and have a person review the result.

Land this

Use AI inside a system that protects the brand.

Participant recap, exercise, and public sources

Recap

Use AI inside a system that protects the brand.

Try it

Use Prompt 5 in the prompt pack. Require editable text, an unchanged approved logo, and a list of any substitutions.

Resources

Use the included NCCC Session 2 Prompt Pack for the reusable prompts and review checklist.

Public sources

- https://www.canva.com/newsroom/news/deep-research-integration-mcp-server/
- https://www.canva.com/newsroom/news/claude-ai-connector/

17. Same process. Two unmistakable brands.

Make AI Look and Sound Like Your Business — Same process. Two unmistakable brands.

Expanded talk track

VoltGuard is a fictional industrial electrical brand aimed at people responsible for operating facilities. Its authority comes from direct language, bold type, dark structure, and a yellow signal color. StarForge represents software, automation, implementation, and guidance through a different visual lineage and language. The two brands may reach some of the same people, but they should not look or sound interchangeable. Their promise, context, and entry point differ. A reusable AI process should preserve that difference rather than flattening everything into the same style. Test the system by imagining a logo swap. If the concept still feels acceptable after swapping only the marks, the surrounding copy and design may not contain enough of either brand. The identity must live in the audience, promise, voice, imagery, layout, and next step, not only in the corner logo.

Land this

A strong process preserves difference.

Participant recap, exercise, and public sources

Recap

A reusable process should preserve difference—not flatten it.

Try it

Use Prompt 4 to compare two brand systems. Check whether swapping only the logos would make either concept feel wrong.

Resources

Use the included NCCC Session 2 Prompt Pack for the reusable prompts and review checklist.

Public sources

- No external source is required for this slide. The examples and exercises are contained in the deck and prompt pack.

18. The click is not the finish line.

Make AI Look and Sound Like Your Business — The click is not the finish line.

Expanded talk track

The ad makes a promise and asks for a next step. The landing page must explain what the service covers, what it cannot prove, what happens next, and which evidence is real. The follow-up should gather the information needed to continue the same process. It should not suddenly become a different offer or a high-pressure pitch. In the VoltGuard example, the ad invites a facility assessment. The destination explains the inspection and its limits. The follow-up gathers facility details and schedules around the customer’s operation. AI can draft each piece from the same approved rules, but people still verify the facts, process, and timing. Choose one real campaign and write down the exact promise made at each stage. Any unexplained change is a point where trust can break.

Land this

Every next step should keep the promise made by the last one.

Participant recap, exercise, and public sources

Recap

Every next step should keep the promise made by the last one.

Try it

Write the exact promise made by an ad, then compare it with the landing page and follow-up step.

Resources

Use the included NCCC Session 2 Prompt Pack for the reusable prompts and review checklist.

Public sources

- No external source is required for this slide. The examples and exercises are contained in the deck and prompt pack.

19. AI can write the post. A marketer runs the campaign.

Make AI Look and Sound Like Your Business — AI can write the post. A marketer runs the campaign.

Expanded talk track

Writing a post is different from operating a campaign. Channel choice, search intent, targeting, budget allocation, retargeting, consent, attribution, and corrective action require tradeoffs. Advertising platforms are deep systems, and their rules change. AI can help an owner arrive with clearer goals, better research questions, a usable brand brief, and draft creative. That can improve collaboration with a marketing professional. It does not transfer responsibility for budget, targeting, claims, measurement, or legal obligations to the model. The distinction resembles software development. Generating pieces is not the same as designing, testing, operating, and maintaining the whole system. Before buying media, identify who owns each decision and how results will be measured.

Land this

AI can prepare pieces. A marketer connects and operates the system.

Participant recap, exercise, and public sources

Recap

AI can prepare pieces. A marketer connects and operates the system.

Try it

Before buying media, identify who owns targeting, budget, review, measurement, and corrective action.

Resources

Use the included NCCC Session 2 Prompt Pack for the reusable prompts and review checklist.

Public sources

- No external source is required for this slide. The examples and exercises are contained in the deck and prompt pack.

20. AI gives ideas. Evidence chooses. Brand filters.

Expanded talk track

AI might suggest phrases such as “commercial electrician near me” or “thermal imaging electrical inspection.” Those are ideas to investigate, not proof of local demand. Google Keyword Planner can provide estimates and forecasts for a location, subject to account access and the tool’s own limitations. Search Console can show the queries that already display or send people to your website, along with impressions, clicks, and click-through rate. A competitor ad proves only that someone ran that message. Without campaign access, we do not know whether it produced useful business. For every idea, separate what was observed from what still needs testing. Then apply the brand filter: even a promising phrase may be wrong if it attracts the wrong customer or encourages a promise the business cannot support.

Land this

Ideas, evidence, and brand fit are different filters.

Participant recap, exercise, and public sources

Recap

Ideas, evidence, and brand fit are three different filters.

Try it

Use Prompt 6. Separate observed facts from ideas to test, then list the evidence still needed for each idea.

Resources

Use the included NCCC Session 2 Prompt Pack for the reusable prompts and review checklist.

Public sources

- https://support.google.com/google-ads/answer/7337243?hl=en
- https://support.google.com/webmasters/answer/7576553?hl=en

21. The evidence is not “AI always wins.”

Make AI Look and Sound Like Your Business — The evidence is not “AI always wins.”

Expanded talk track

The research does not support a simple claim that AI advertising always wins or always fails. A survey in which business leaders report that social media feels effective does not establish measured return on ad spend. Consumer studies may describe a particular age group, time period, or experimental setup rather than all consumers. The cited sources point toward a more practical conclusion. Execution and relevance matter. Obvious or distracting AI use can hurt the message, while seamless use may work. Consumers and advertising executives may also judge the same material differently. Before repeating a statistic, record who was surveyed, when the study occurred, and what the number actually measures. The studies on this slide use different samples and methods, so their percentages should not be combined into a trend line.

Land this

The production method is not the selling point. Relevance and execution decide whether the work helps.

Participant recap, exercise, and public sources

Recap

Execution and relevance matter more than announcing which tool made it.

Try it

Read the study methods before repeating a statistic. Record who was surveyed, when, and what the number actually measures.

Resources

Use the included NCCC Session 2 Prompt Pack for the reusable prompts and review checklist.

Public sources

- https://quickbooks.intuit.com/r/small-business-data/advertising-trends-2025/
- https://www.kantar.com/Inspiration/Advertising-Media/Rethinking-AI-generated-advertising
- https://www.iab.com/insights/the-ai-gap-widens/

22. Before “publish,” run five checks.

Make AI Look and Sound Like Your Business — Before “publish,” run five checks.

Expanded talk track

A polished draft can still promise a service the business does not offer, quote a review that is not real, imply a result the evidence cannot support, or use an image that misrepresents the product. Before publishing, run five checks. Is the message for the right customer and offer? Does it look and sound like this business? Can we prove every claim, review, number, and image? Are the logo, text, and image separate and usable in the final format? Did the right person approve the piece before publication or spending? Keep private customer data out of unapproved tools. Review the final format, not only the draft text. An assistant being able to publish or spend does not mean anyone authorized it.

Land this

Capability is not authority. Human approval remains the gate.

Participant recap, exercise, and public sources

Recap

Capability is not authority. Human approval remains the gate.

Try it

Run Prompt 7 against one draft. Resolve every must-fix issue before anyone publishes or spends money.

Resources

Use the included NCCC Session 2 Prompt Pack for the reusable prompts and review checklist.

Public sources

- No external source is required for this slide. The examples and exercises are contained in the deck and prompt pack.

23. Not seven separate prompts. One business.

Make AI Look and Sound Like Your Business — Not seven separate prompts. One business.

Expanded talk track

The audience and promise that improved the first post should also shape the offer, landing page, proof, follow-up, and measurement. We are not asking AI to invent a new company whenever we need another piece of marketing. Save one approved brief and reuse it across the campaign. Record the audience, promise, voice, visual rules, known facts, prohibited claims, and next step. When measurement shows that one message worked better for a particular audience, feed that learning back into the brief without erasing the identity. Change one useful variable at a time when testing. If the visual style, offer, audience, copy, and destination all change together, the result teaches very little. Useful measurements might include relevant inquiries, booked assessments, and the questions customers ask, not merely likes.

Land this

One business system should guide the whole loop.

Participant recap, exercise, and public sources

Recap

We are not asking AI to invent a new business every time we need a post.

Try it

Save one approved brand brief and reuse it across the campaign. Change one variable at a time when testing.

Resources

Use the included NCCC Session 2 Prompt Pack for the reusable prompts and review checklist.

Public sources

- No external source is required for this slide. The examples and exercises are contained in the deck and prompt pack.

24. Make it unmistakably yours.

Make AI Look and Sound Like Your Business — Make it unmistakably yours.

Expanded talk track

You do not need an art degree to begin. You need to notice what you like, explain why, and commit to a few consistent choices. Write one page that names the audience, promise, voice, visual rules, verified facts, factual boundaries, and next step. Use that page for one useful post. Review the draft. Keep the corrections. If the result feels generic, ask which business decision is missing rather than asking the model to make it more exciting. If the draft contains a claim, find the evidence or remove it. The food truck, dentist, and salon from the beginning never needed the same costume. Their customers, situations, and promises were already different. The brand brief gives the AI a way to respect those differences.

Land this

Your brand gives AI the rules. AI helps with the draft. You still decide.

Participant recap, exercise, and public sources

Recap

Your brand gives AI the rules. AI helps with the draft. You still decide.

Try it

Write your one-page brand brief, use it for one useful post, review the result, and save your corrections.

Resources

Use the included NCCC Session 2 Prompt Pack for the reusable prompts and review checklist.

Public sources

- No external source is required for this slide. The examples and exercises are contained in the deck and prompt pack.

25. Questions—and one useful next step.

Make AI Look and Sound Like Your Business — Questions—and one useful next step.

Expanded talk track

Use the remaining time for questions that help the room. A detailed account setup or a business-specific campaign may need a separate conversation. For this session, identify the next useful step and avoid turning the close into a private implementation consultation. Choose one action you can finish this week. Define clearer voice rules. Build the one-page brand brief. Review one campaign draft with the five checks. One finished improvement gives you evidence to use in the next iteration. Remember that VoltGuard is fictional and the prepared examples are teaching samples. The prompt pack contains the reusable workflows, but the facts and approvals must come from your real business.

Land this

One useful next step beats seven unfinished experiments.

Participant recap, exercise, and public sources

Recap

One useful next step beats seven unfinished experiments.

Try it

Choose one next step you can finish this week: clearer voice rules, a better brief, or one reviewed campaign draft.

Resources

Use the included NCCC Session 2 Prompt Pack for the reusable prompts and review checklist.

Public sources

- No external source is required for this slide. The examples and exercises are contained in the deck and prompt pack.

26. Sources, prompts, and notes

Make AI Look and Sound Like Your Business — Sources, prompts, and notes

Expanded talk track

The sources are here so the claims remain inspectable after the session. The participant prompt pack contains the thin request, the branded request, the voice-mapping workflow, the two-brand comparison, the editable-design workflow, the research prompt, and the pre-publication audit. Use the public research links to check the underlying article, study methods, and current product documentation. Platform features and account access can change, so documentation should be treated as a current reference to verify rather than a permanent promise. The fictional VoltGuard example is a practice environment, not a source of real credentials, response times, reviews, savings, or safety records. When you reuse a prompt, replace the fictional details with verified facts from your business and keep the factual boundaries intact.

Land this

Inspectable sources make the system easier to trust and improve.

Participant recap, exercise, and public sources

Recap

Sources are here so the claims are inspectable.

Try it

Keep the included prompt pack beside this deck. Use the public research links when you need to inspect the underlying evidence.

Resources

The included NCCC Session 2 Prompt Pack contains all seven reusable prompts and the source list.

Public sources

- https://seths.blog/2006/04/ode_how_to_tell/
- https://www.kantar.com/Inspiration/Advertising-Media/Rethinking-AI-generated-advertising
- https://quickbooks.intuit.com/r/small-business-data/advertising-trends-2025/
- https://www.iab.com/insights/the-ai-gap-widens/
- https://www.canva.com/newsroom/news/deep-research-integration-mcp-server/
- https://www.canva.com/newsroom/news/claude-ai-connector/
- https://support.google.com/google-ads/answer/7337243?hl=en
- https://support.google.com/webmasters/answer/7576553?hl=en