Eric Rubin

I wrote 48 competitor alternatives articles last year. If you work in SaaS content, you've probably written dozens too and you know they take a hell of a lot of time to create. Research every tool. Verify every pricing page. Read G2 reviews until your eyes glaze. Check for lawsuits, controversies, community backlash. Write it up in a voice that doesn't sound like a feature comparison spreadsheet. 8+ hours per article, easy. I built a skill that cuts that to 2-3 hours. Before you come at me saying why bother writing content if you're using AI: this isn't a one-shot article generator. It's a human-in-the-loop workflow where the agent becomes your research assistant. You still gotta know your competitors like the back of your hand. You just give it direction. You decide the angle, pick the tools, shape the framing. It does the grunt work, stops at checkpoints, and waits for your approval before moving on. Here's what actually happens when you run it: It proposes which tools to cover and why people are leaving the title tool. You approve or adjust. It goes deep on research from primary sources only: official websites, G2, Reddit, pricing pages. It digs up lawsuits and controversies. It comes back with a summary and source URLs for every major claim. You review, correct, approve. Then it writes and drops a formatted .docx in your output folder. The articles get better over time too. Drop your past articles in the folder and it picks up your voice, avoids repeating structure, varies the shape across pieces so Google doesn't flag them as templated. Setup is three files: the skill, an AGENTS.md template where you fill in your product's positioning, pricing, limitations, and voice, and an outputs folder where finished articles land. If you write comparison content for your product, DM me or comment and I'll send you the setup. #claudeskill

37

56 Comments

The marketer's job title didn't change. The job description did. Most don't know it yet. I've been deep in Claude Cowork for the past few weeks. I'm fascinated by it, and I've been reading everything I can find about how other marketers are using it. Here's what I'm seeing: 𝗙𝗿𝗼𝗺 𝗣𝗿𝗼𝗺𝗽𝘁𝗶𝗻𝗴 𝘁𝗼 𝗣𝗿𝗼𝗴𝗿𝗮𝗺𝗺𝗶𝗻𝗴 I've stopped writing prompts. I'm building "Skills" now - structured markdown files that encode my methodology, brand voice, and operational playbooks. One file. Reusable forever. The agent references it on every task without re-prompting. I'm using the "Skill Creator" - a meta Skill that helps you build other Skills.. 𝗧𝗵𝗲 𝗡𝘂𝗺𝗯𝗲𝗿𝘀 𝗜'𝗺 𝗦𝗲𝗲𝗶𝗻𝗴 → 4x content output at 25% of previous cost (Marketing Agent Blog) → 127% faster content creation while maintaining 89% quality (Stanford GSB) → 40-60% reduction in project completion time via sub-agent coordination (Claude Help Center) → Tasks that took 2.5 hours now take under 30 minutes (Anthropic research) But this stat stopped me: 27% of Claude-assisted work consists of tasks that would never have been done otherwise (Anthropic). The exploratory research, the "nice-to-have" dashboards, the competitive deep dives that always got deprioritized. 𝗧𝗵𝗲 𝗡𝗲𝘄 𝗢𝗿𝗴 𝗖𝗵𝗮𝗿𝘁 Some marketers are running entire operations on autopilot. A "CMO" agent reads a weekly plan, spawns specialist sub-agents (content writer, social marketer, newsletter editor), and they execute in parallel. Cost: $20-50/day in Claude API usage (Snow Lee on Medium). 𝗧𝗵𝗲 𝗖𝗮𝘃𝗲𝗮𝘁 97% of marketers still edit AI-generated copy before publishing. The productivity gains come from accelerating the first 80% - research, drafting, formatting. Human judgment handles the rest. Without that human layer, you get generic slop. Repetitive patterns. A uniform voice. Content fatigue. I feel that the role is permanently splitting: AI handles execution at scale. Humans handle taste, strategy, and brand.

19

3 Comments