What This Covers
For ten months, one non-technical marketer ran all of Anthropic's paid search, paid social, app store optimization, email marketing, and SEO. This post breaks down his exact workflow, the results he produced, and what this model means for small business owners building lean marketing systems.
What Happened at Anthropic's Growth Marketing Team?
Starting in Q2 2025, Austin Lau was Anthropic's entire growth marketing department. Not the lead of a team. The team.
He managed paid search campaigns on Google, paid social on Meta, app store optimization across platforms, email marketing, and SEO. He did this without writing a single line of code in the traditional sense.
Instead, he used Claude Code, Anthropic's own AI tool, to build custom automation workflows in plain language.
The story circulated widely in early 2026 after Anthropic published a case study documenting his approach. It spread because it confirmed something many people had been quietly observing: the ratio of marketing output to headcount is shifting fast.
Who Is Austin Lau and Why Does His Story Matter?
Austin Lau is a growth marketer by background, not a developer. He had no prior coding experience before building his AI-powered workflows. What changed was access to Claude Code, a tool that lets non-technical users build and run automated systems by describing what they need in plain language.
His approach is not theoretical. It is documented, reproducible, and built on tools that are already available to any business. That is what makes it significant.
As Lau put it: "All you need to know is how to explain your challenge and what you're trying to solve in a very clear manner."
What Did His Workflow Actually Look Like?
Lau built two primary systems within one week of starting with Claude Code, then expanded from there.
The /rsa command for Google Ads
He created a custom slash command called /rsa for responsive search ads. Typing it into Claude Code triggered a structured workflow: the system requested his campaign data, existing ad copy, and target keywords. It cross-referenced those inputs against a set of instructions he called Agent Skills.
Those Agent Skills contained Anthropic's brand voice guidelines, product accuracy rules, and Google Ads best practices. Two sub-agents then handled output. One wrote headlines. One wrote descriptions. Each operated within the correct character limits automatically.
The Figma plugin for ad creative
He built a Figma plugin that generated ad creative variations with one click. For every format and aspect ratio a campaign required, the plugin produced variations automatically, applying brand guidelines without manual resizing or redesigning.
Each large creative update that previously took 30 minutes now took 30 seconds.
The CSV performance loop
For ad performance analysis, he exported campaign data as CSV files and fed them into Claude Code. The AI identified underperforming ads, generated replacement copy, and produced upload-ready files he could push directly back to the platform.
The cycle that once took hours became a repeatable routine.
What Were the Measurable Results?
Anthropic published the results from their internal case study. The numbers are specific:
From 30 minutes to 30 seconds per ad set
Increase in variations from a single workflow
From 2 hours to 15 minutes
Across the broader Anthropic marketing organization, the pattern repeated:
- Influencer marketing saved 100+ hours per month
- Customer marketing reduced case study drafting from 2.5 hours to 30 minutes
- Digital marketing logged a 5x productivity increase year-over-year
- Product marketing saved 5 to 10 hours per product launch
Is This a One-Off Story or a Pattern?
Anthropic's growth team has since expanded. The story is not that one person can do this forever. The story is what became possible in the first place.
Across the industry, agencies are restructuring around fewer humans and more AI operators. Freelancers are running campaigns that previously required full teams. Job postings increasingly describe roles like "AI marketing operator" that combine strategic judgment with workflow-building skills.
The tools that made this possible at Anthropic are the same tools available to any business. The gap is not access. It is knowing how to structure the workflow.
What Does This Mean for Small Business Marketing?
The takeaway is not "replace your team with AI." The takeaway is that the team size required to do effective marketing has changed.
A single person with well-built AI workflows can now produce consistent output across:
- Paid search copywriting and iteration cycles
- Ad creative production across multiple formats
- Performance analysis and copy refresh
- Email drafts and subject line testing
- SEO content and keyword research
This does not happen automatically. It requires building the workflow, testing it, refining it, and documenting the brand rules that guide AI output. That investment compounds. Teams that build it now accumulate an advantage that grows over time.
Where Is This Heading?
Several patterns are already visible across the industry:
- Enterprise marketing teams are reporting the same 5 to 10x productivity multipliers Anthropic documented
- Small agencies are moving toward a model of two to three people running work that previously required ten
- The expectation for what a solo marketing hire can produce is rising across all sectors
Within two to three years, businesses that have built AI-assisted marketing workflows will have a meaningful structural advantage over those still operating on a fully manual model. The tooling is already available. The gap is in who builds the systems first.
How Do You Start Building This Kind of Workflow?
Start narrow. Pick one high-repetition marketing task. Writing social captions, refreshing Google Ads copy, and generating email subject line variants are all good entry points.
Build one repeatable AI-assisted process around that task. Document what works. That documentation becomes the equivalent of Lau's Agent Skills. It is the brand reference material that makes AI output consistent.
Expand from there. Each workflow you systematize frees time for the strategic decisions AI cannot make for you. Strategy, positioning, and relationship-building remain human work. The volume work does not have to be.
Tools Referenced in This Post
- Claude / Claude Code (claude.ai, code.claude.com) — the AI used by Anthropic's growth team to build all workflows
- Figma (figma.com) — used for ad creative automation via a custom plugin
- Google Ads Editor — for bulk-uploading AI-generated copy via CSV
- Notion or Google Docs — for storing brand voice guidelines that AI references
AVUSEN helps businesses build smarter systems for marketing, design, and digital presence.