This month, I joined my first hackathon — something I never thought I'd do, as I never considered myself a hacker. The advent of vibe coding changed that, as it has allowed me to launch dozens of marketing tools and pet projects. 

I shared thoughts about the hackathon itself on LinkedIn, including how I'd go about it differently. But here, I wanted to talk about the nine-step process for how to approach vibe coding, lest you think it's more daunting than it is. The hardest step is the second one, but once you start going through the process, all of these steps become easier.

1) Be open to ideas coming from anywhere. It's hard to find ideas without being open to the possibility for them to come to you. You need to put up your antenna. On a dating app, a conversation about Toll House cookies didn't lead to any romance, but it did lead to me launching the recipe site backofthebag.com. 

2) Seize a moment where you wish something existed. In the case of the recipe site, I thought it would be great to have an outlet that is only focused on recipes printed on mass market consumer goods' packaging. A lot of my favorite things to cook and bake are from such sources, including Hershey's hot cocoa and, of course, Rice Krispie treats.

3) Research if the idea exists. There is hardly a need for more recipe sites, but I fired up Gemini and asked if it could find anything like this. If you're launching something that is core to your brand and could become a long-term investment, spend a lot of time on this step. But if it's a passion project you're exploring on a whim, it's safe to just make sure you're not outright copying something else. Don't let “perfect” be the enemy of “good enough.“

3a) If the idea already exists, see if it exists in the way you wish it did. Maybe such a recipe site exists, but it's owned by a single packaged goods conglomerate. Maybe a British version exists but not an American one. Maybe it exists as a blog but without capabilities like saving recipes. Find your niche within a niche.

3b) If there are potential legal implications, do a cursory review. When launching a pet project, you (probably) don't need a lawyer and can see if you need to put some disclaimers on your site. Alternatively, you might find out right away that launching it is a legal minefield and you should think of something else. AI shouldn't replace lawyers, but you can get a sense of whether you might need one later. 

4) Start building it. Don't wait too long to start prompting your vibe-coding app of choice. It could be Claude Code or ChatGPT Codex, or a standalone app like Lovable or Replit. The tool matters less than your idea, clarity, and capacity to iterate. Get the first prompt in. Refine. Adapt. Play.

5) Make it feel real. I buy domain names for every project I attempt to launch. Often, I'll switch this step with the previous one, as having a great domain makes me more excited about the idea. I had to build my site for job seekers when I got the domain imfindable.com. I prefer .com domains for their credibility, memorability, and price; they're way cheaper than .ai domains, especially when you renew them. Beyond domains, I set up each site on Google Search Console. It takes all of a minute to do so, and GSC integrates with major domain providers like GoDaddy, along with vibe-coding apps like Lovable and Base44.

6) Make it functional. See how far you can take it to get to the minimum viable product (the closest you and I will get to an MVP unless we hang with Jalen Brunson). That might mean setting up email forms and subscriptions, or e-commerce functionality through Stripe or Paddle. The major vibe-coding platforms have integrations built in and may direct you to their preferred options; use their defaults to get started. See if you can get one aspect of what you want to build to work. I can't tell you how satisfying that is.

7) Publish it! I tend to publish as quickly as I can so I can test it in the wild and really see what works or not. You may want to spend longer in the editor or on the staging site first.

8) Test and refine. If you build something that looks and works exactly as you want it to, congrats, but few projects' first versions (or first 20 versions) go as planned. Beyond all kinds of things breaking with vibe coding, you'll likely keep having ideas for improvements. Stuck, or need more input? Use the Gemini extension in Chrome and ask it for ideas while you go through your site. I've done some invaluable debugging and enhancements that way.

9) Share it. Maybe you keep it quieter, like adding it to a portfolio page of yours or your CV. Perhaps you want to share it with a few friends to test it. If you're proud of it, it works, and you want to get it attention, then tap into your inner or outer marketer and get it out there. There's also no better way to get helpful suggestions for improvement than by sharing it with others. 

Maybe the process ends there. You probably learned something. Congrats! But if you learned a lot and you're proud of where you got so far, you might keep this process going indefinitely.

That's the real danger of vibe coding — not that you'll waste your time on a failed project, but that you'll become hooked. Bet you can't vibe just one.

— David Berkowitz, Chief Community Officer, Marketecture Media

Who: CTV Advertisers, Brand Marketers, Media Buyers, SMB Marketing Teams, Agencies

What: Disney is set to launch a beta AI ad creation tool that can generate scripts, video, and music from existing brand assets. The goal is to make connected TV advertising more accessible, especially for smaller advertisers that lack large creative teams or production budgets.

Why it matters: AI-generated creative is moving from social platforms into premium TV inventory. Brands that master AI-assisted video production early may gain speed and cost advantages in CTV.

(Business Insider)

Who: Agency CEOs, Account Directors, Operations Managers, Creative Tech Leads

What: Indie agency January Digital shared details on its internal AI strategy, highlighting a decentralized approach where all employees received custom platform access to build their own workflow tools. Despite these internal successes, the agency noted growing tension around client management, as client organizations vary wildly in their readiness for AI adoption and frequently expect steep fee discounts before real efficiency gains have materialized.

Why it matters: True agency AI adoption requires a cultural shift that treats junior staff as tool builders rather than automation victims, but agencies must heavily manage client expectations to prevent artificial intelligence from shrinking profit margins before the tech actually matures.

(Digiday)

Who: E-commerce Directors, Retail CMOs, Digital Marketers, Growth Managers

What: Consumers who find retail websites through large language models are staying longer and spending more money than traditional visitors. Adobe Analytics data indicates that traffic coming from conversational AI platforms generated 53% more revenue per visit and resulted in a 54% higher conversion rate compared to non-AI traffic sources.

Why it matters: Shoppers coming from AI engines arrive with incredibly high purchase intent because they have already completed their product research and comparisons inside the LLM conversational interface. Brands must optimize their web content to be easily indexed and recommended by AI assistants.

(Reuters)

Liking AI Brief? The Refresh is a weekly recap of all the most important advertising news, written by AdTechGod.

Thanks for reading this week’s AI BRIEF by Marketecture. 

Marketecture Media is committed to bringing you the latest news, opinions, and community updates. 

If you are interested in advertising on this newsletter, our podcasts, The Advertising Forum, or in our Slack community, you can contact us here.