How One Marketer with an Agent Stack Replaces a Department

How One Marketer with an Agent Stack Replaces a Department

How One Marketer with an Agent Stack Replaces a Department

WRITTEN BY
Jon Kruzeniski

A single growth marketer at Anthropic was running what previously required a team spread across multiple individual roles, using Claude Code to coordinate execution that used to demand five to ten specialists across different functions.

That dynamic is not limited to marketing. On the same day this was written, Coinbase CEO Brian Armstrong announced a 14 percent reduction in workforce, citing AI directly. Over the past year, he said, engineers at Coinbase shipped in days what used to take a team weeks. The company is now reorganizing around what Armstrong calls AI-native pods, including one-person teams combining engineering, design, and product responsibilities. The pace of what is possible with a small focused team, he said, has changed dramatically and is accelerating every day.

The same shift is happening in marketing.

What Those Five to Ten People Actually Were

Open Indeed right now and search for marketing roles. What comes back is a map of every function a modern marketing department needs to cover, each posted as its own role:

Social Media Manager. Paid Acquisition Manager. Content Marketing Manager. Email Marketing Manager. Growth Marketing Manager. Demand Generation Manager. Lifecycle Marketing Manager. Influencer Marketing Manager.

Each one a full-time job. Each one mapping directly to what an agent stack now handles at the execution layer.

And that list does not include the crossover functions that have always bled into marketing work. Design, product, engineering, and data analytics have historically required collaboration across teams or additional headcount to support campaigns, landing pages, creative production, and attribution. Agents now cover that overlap too.

Each of these is not a single agent but a system. A paid agent coordinates subagents handling creative production, copywriting, and performance data. A content agent coordinates subagents handling research, drafting, and distribution. The pattern holds across every function. What looks like one hire on an org chart is one agent stack coordinating several.

The agent stack does not eliminate these functions. It absorbs the execution load that made each one a full-time hire.

What This Cost Before

Seed stage companies typically started with a single Marketing Manager covering the full function. Social media, email, content, paid, and influencer, often supported by writers, agencies, and contractors to extend the output. As the company grew and built out a marketing leadership team, dedicated specialized roles got added to handle the increasing execution load across each channel.

As a marketing team evolved beyond the initial generalist hire, it was not unusual to see dedicated roles covering each major channel. A Digital Marketing Manager handling paid acquisition. A Social Media Manager owning organic channels. A Content Marketing Manager running the blog, email copy and SEO. A Lifecycle and Email Marketing Manager owning email workflows, retention and post-purchase flows. Larger teams added headcount within each of those functions and layered in outside agencies and specialist firms on top.

At scale it went further. A prominent head of growth at a well-known Bay Area startup once told me he had 16 people reporting to him, including four Influencer Marketing Managers alone. A lot has changed since that conversation, but it gives you a sense of the headcount marketing teams could reach in the pre-AI era.

The data confirmed it was not an outlier. A SaaS startup with $1 to $5 million in revenue carried a marketing team of around three people. By the time a company reached $20 to $50 million that team had tripled. The industry scaled headcount because it was the primary way to scale execution.

What One Person Can Run Now

The shift the Anthropic story describes is not about one shortcut on one task. It is about the execution layer changing.

To be precise about what that means: agents handle execution. The workflows, the creative production, the distribution, the sequencing, the reporting. What they do not replace is the strategy behind the system. Positioning, segmentation logic, experiment design, reading weak signals before they appear in the data, understanding why a cohort behaves a certain way, that judgment still sits with a human. The agent stack is not a replacement for marketing intelligence. It is infrastructure that eliminates the headcount previously required to act on it.

A single marketer running an agent stack can now cover what previously required hiring across multiple roles. A content agent producing consistent output across formats and channels. A paid agent generating, testing, and iterating on creative across platforms. A social agent maintaining presence and engagement. An email agent managing sequences and sends based on behavioral triggers. A lifecycle agent executing post-purchase flows designed by the strategist above it. A demand generation agent coordinating distribution across top of funnel programs. A data agent surfacing reporting and performance signals fast enough to act on.

Each agent a system in itself. The marketer above the stack the strategist who designs what the system runs.

The Structure Has Changed

The execution load that defined how marketing teams were built has a different answer now. The functions still need to get done. What has changed is that one marketer with the right agent stack can now absorb the execution that previously required hiring across all of them.

The Anthropic marketer was running what used to require five to ten people. The question is whether you build the team the industry built before, or build the system that makes that team unnecessary.

How One Marketer with an Agent Stack Replaces a Department

The roles are still real. The headcount requirement is not.

How AI Agents Work at Every Stage of the Funnel

How a connected agent stack collapses growth silos and builds compounding intelligence across the full funnel.

The Economics of Timing in Customer Acquisition

How Payback Periods Shape Unit Economics

Growth Engines That Endure: When Growth Becomes Infrastructure

The endpoint of growth is not acceleration. It is continuity. Growth becomes infrastructure when it continues to function without constant intervention.

Margin of Safety, Attribution & Growth Engines That Scale

The concept of a “margin of safety” — the idea that decisions should be structured to absorb uncertainty and error — was first formally introduced by Ben Graham and David Dodd in the 1930s, in their landmark book Security Analysis. Seth Klarman’s modern book Margin of Safety popularized the principle for contemporary investors, emphasizing disciplined decision-making and risk management.

Attribution: Preserving Causality as Systems Scale

Most growth metrics describe outcomes after they occur. Velocity describes how a system behaves while it is operating. At Kruzeniski Digital, velocity is observed as time compression — the duration between input, feedback, and adjustment.

Growth Velocity: Time as a Measure of System Health

Most growth metrics describe outcomes after they occur. Velocity describes how a system behaves while it is operating. At Kruzeniski Digital, velocity is observed as time compression — the duration between input, feedback, and adjustment.

Margin of Safety: Designing Growth That Survives Error

In investing, margin of safety is not about prediction. It is about durability. Growth systems operate under similar conditions. Assumptions will fail. Performance will vary. Conditions will change. The relevant question is not whether a system works under ideal circumstances, but whether it continues to function when those circumstances shift.