Andrej Karpathy coined “vibe coding” in February 2025 — give in to the vibes, let the AI write the code, don’t even read the diffs. A year later, Greg Isenberg declared the same 20x acceleration is hitting marketing: “It’s called VIBE MARKETING.” Searches for the term are up 686% year-over-year. MarTech published a “Vibe Marketing Manifesto.” Morning Brew is covering it. Nearly half of the Fortune 500 are now using vibe marketing approaches in some form. Startups are hiring “Vibe Marketers” at salaries up to $1 million. The movement has arrived. And it has a ceiling.

Key Takeaway: Vibe marketing — running your marketing through Claude Code sessions, n8n workflows, and DIY agent chains — works brilliantly for technical founders who enjoy building systems. It breaks down for the 95% of founders who want marketing outcomes, not marketing infrastructure. The gap between “I can build this” and “this runs while I sleep” is where most vibe marketing setups die. That gap is the product opportunity.

What Vibe Marketing Actually Is

The term comes from Karpathy’s original tweet: “There’s a new kind of coding I call ‘vibe coding,’ where you fully give in to the vibes, embrace exponentials, and forget that the code even exists.”

Applied to marketing, it means: use AI tools to do in hours what used to take teams weeks. Greg Isenberg’s framing captures the vision:

Marketing calendars auto-generated by AI agents
A single marketer with AI agents testing dozens of angles in real time
Campaign creation compressed from 8-week cycles to 2-day sprints
One person managing what used to require a 10+ person team

The ecosystem is already here. Stormy AI publishes playbooks on running Google and Meta Ads through Claude Code. ClaudeKit Marketing ships 32 AI agents and 68 skills for $99. MKT1 Newsletter profiles real marketing teams building custom Claude Code workflows. SE Ranking built an MCP server that pipes live SEO data directly into Claude sessions.

This isn’t hype. People are genuinely doing this. @DePaulaAllyson ran a 4-day vibe marketing experiment wiring up OpenClaw, Meta Ads, and Microsoft Clarity — her X thread documenting the process got 26 likes, which in the vibe marketing community counts as a hit. It works. The question is: for whom, and for how long?

The Five Walls

Every vibe marketing setup hits the same walls. Not immediately — these systems work great in week one. The problems compound over time.

Wall 1: It Requires a Developer

Vibe coding works because Karpathy is a world-class ML researcher who can evaluate whether the AI’s code is correct. Vibe marketing works because the people writing about it are technical founders who can wire up MCP servers, debug API authentication, and troubleshoot n8n workflows at 2 AM.

That’s not most founders. Most founders started a company to solve a problem for their customers, not to become their own marketing DevOps team. The moment your Claude Code session throws a Mailgun authentication error or your n8n webhook silently stops firing, you’re not doing marketing anymore — you’re doing infrastructure maintenance.

Digital Applied reports that “agencies using Claude Code sub-agents report a 75% reduction in time spent on repetitive tasks.” Agencies. Not solo founders. People who already have the technical foundation to build and maintain these workflows.

Wall 2: No Persistence

Every Claude Code session starts from scratch. Your AI doesn’t remember that last Tuesday’s LinkedIn post about pricing strategy got 3x the engagement of the product update. It doesn’t know that your audience in Southeast Asia responds better to case studies than thought leadership. It doesn’t carry forward the lesson that your open rates drop on Fridays.

You can paste context into every session. You can maintain a CLAUDE.md file with brand guidelines. But that’s not learning — that’s manual memory management. The human becomes the persistence layer, which means the human is still the bottleneck.

And the tools themselves don’t persist either. OpenClaw racked up 60,000 GitHub stars in 72 hours and its creator was acquired by OpenAI — the darling of the vibe marketing toolchain, absorbed before most people finished their first workflow. Weeks later, Nous Research launched Hermes Agent with persistent memory, and Greg Isenberg was already asking “Is Hermes Agent the new OpenClaw?” The underlying infrastructure churns faster than anyone can build stable processes on top of it. Your carefully assembled vibe marketing stack from January is already legacy by March.

Real marketing compounds. Every campaign teaches you something about your audience. Every A/B test narrows the search space. Every customer conversation refines your positioning. A system that forgets everything between sessions can’t compound. It can execute. It can’t learn.

Wall 3: No Trust Model

Vibe marketing operates in binary: either you review everything before it goes out, or you let the AI post directly. There’s no middle ground — no “this type of content can go live automatically, but that type needs my approval.” No escalation path for edge cases. No cultural calendar that knows not to run a cheerful promotion during a national tragedy.

Digiday reported that CMOs are adopting AI more slowly than it’s evolving — not because of capability concerns, but because of governance. Who’s accountable when the AI agent publishes something tone-deaf? In a vibe marketing setup, the answer is always “you” — because you’re the only human in the loop. That doesn’t scale.

Wall 4: No Orchestration

A real marketing operation isn’t one channel. It’s blog content that feeds social posts that drive email signups that trigger nurture sequences that book demo calls. Each step depends on the previous one. The timing matters. The message consistency matters.

In a vibe marketing setup, you orchestrate this manually. You run one Claude Code session to write the blog post. Another to create social variants. Another to draft the email. You copy-paste between tools. You check that the messaging is consistent. You schedule everything in the right order.

This is what Grippi Media’s analysis found: “Vibe workflows can handle simple tasks. But once you add more tools and steps, things constantly break and cost you more time than it saves.” Each platform has different data formats, rate limits, and authentication methods.

You haven’t eliminated the coordination problem. You’ve just moved it from a team of humans to a team of one human juggling multiple AI sessions.

Wall 5: The Economics Don’t Work for Persistence

Running Claude or GPT-4 costs money per token. A quick session to draft a blog post or audit your ad spend is cheap. A persistent agent that monitors your brand mentions, tracks competitor moves, analyzes your email performance, and proactively suggests campaign adjustments every hour of every day — that’s a fundamentally different cost structure.

This is why every vibe marketing tool is session-based. You open it when you need it. You close it when you’re done. But marketing doesn’t stop when you close your laptop. Your competitors are publishing. Your customers are talking about you. Your campaigns are running. The market is shifting.

The gap between “I use AI when I sit down to do marketing” and “AI does marketing while I build my product” is the gap between a tool and a teammate.

The Cursor Analogy (and Why It Breaks)

Greg Isenberg calls vibe marketing “what Cursor did for coding, but for marketing.” It’s a sharp analogy. It’s also revealing.

Cursor made individual developers more productive. It didn’t replace engineering teams. It didn’t eliminate the need for architecture decisions, code review, deployment pipelines, or production monitoring. It accelerated the typing. The thinking, the coordination, the institutional knowledge — those still require humans and systems.

The same is true for vibe marketing tools. They accelerate the doing. They don’t solve the knowing — what to do, when to do it, whether it’s working, and what to try next.

We’ve written before about the copilot ceiling: tools that make you faster at tasks you already know how to do. The ceiling is that you still need to know what to do. For a solo founder who isn’t a marketer, “faster execution” isn’t the bottleneck. Knowing what to execute is.

What Comes After Vibe Marketing

Every technology wave follows the same arc. First, the DIY phase — enthusiasts wire together tools and share workflows. Then, the product phase — someone packages the best practices into something that just works.

We saw this in coding:

DIY phase: Copy-paste from ChatGPT into your editor
Product phase: Cursor, GitHub Copilot, Windsurf — AI embedded in the workflow

We saw this in data:

DIY phase: SQL queries in notebooks, custom Python scripts
Product phase: dbt, Looker, Amplitude — the messy middle automated away

Marketing is in the DIY phase right now. Claude Code sessions. n8n workflows. MCP servers. ClaudeKit bundles. It works for the people building it. It won’t become the way most founders do marketing.

The product phase looks different. Not “here’s an AI tool, go build your marketing” but “here’s an AI that already knows your brand, already connected to your channels, and is already running campaigns you can approve or adjust.” Not a copilot you drive. A colleague that drives while you navigate.

The difference between vibe marketing and what comes next is the difference between assembling IKEA furniture and hiring a contractor. Both get you a kitchen. One requires you to be the builder.

The Real Question

Christopher Penn noted that vibe marketing sounds “interesting at best and sleazy at worst” to many marketers. The Content Marketing Institute argues for “slow-motion marketing” as the antidote. Grippi Media says flatly: “Vibe Marketing Isn’t the Future of Marketing.”

They’re all partially right. Vibe marketing isn’t sleazy — it’s genuinely useful. It’s also not the future — it’s the present transition state. The scaffolding before the building.

The real question isn’t whether to use AI for marketing. That’s settled. The question is whether you want to be the one building and maintaining the AI marketing system, or whether you want one that works on its own.

For the technical founder who loves tinkering with Claude Code at midnight — vibe marketing is perfect. Ship it.

For everyone else — the founder who needs marketing to happen while they’re on a customer call, the solo marketer juggling five channels at a seven-person company, the startup that can’t afford to hire but can’t afford to stop growing — what comes next is the thing that actually scales.

Not faster tools. A Brand Parent that carries persistent concern for your brand, learns from every interaction, and acts on your behalf — with guardrails, with cultural awareness, and without requiring you to be a developer to use it.

The vibes were a good start. Now build the building.

References

Andrej Karpathy. “Vibe Coding” (original tweet), February 2025.
Greg Isenberg. “Vibe Marketing Is Here” (tweet), March 2025.
DemandSage. Vibe Marketing Statistics — Growth & Usage Trends, 2025.
Stormy AI. The Rise of Vibe Marketing with Claude Code, 2025.
Digital Applied. Claude Code Subagents: Digital Marketing Guide, 2025.
MKT1 Newsletter. What 4 Gen Marketers Are Building with Claude Code, 2025.
SE Ranking. MCP Server — SEO Data API, 2025.
Digiday. “Agentic with a small a”: CMOs are adopting AI more slowly than it’s evolving, 2025.
Grippi Media. Vibe Marketing Isn’t the Future of Marketing, 2025.
Backlinko. Vibe Marketing: Hype, Reality, and Real Case Studies, 2025.
Christopher Penn. What Is Vibe Marketing?, March 2025.
Content Marketing Institute. Slow-Motion Marketing is the Antidote to Vibe Marketing, 2025.
ClaudeKit. ClaudeKit Marketing — AI Research, Sales & Growth Automation, 2025.
MarTech. The Vibe Marketing Manifesto, 2026.
@DePaulaAllyson. 4-Day Vibe Marketing Experiment (X thread), 2026.
Morning Brew. Vibe Marketing coverage, 2026.

This is part of our series on the Brand Parent thesis — why marketing needs persistent AI agents, not just faster tools. Previously: The Copilot Ceiling, Slop vs Craft, a16z Says Copilots Are a Dead End.