How I AI: Alex Lieberman's 6-Step Claude Workflow to Beat AI Slop

When I talk to founders, especially technical ones, I tell them they have to do the one thing they dread most: climb Cringe Mountain. You have to post on X. You have to post on LinkedIn. You have to talk about your work and your business, even when it feels unnatural. The old marketing channels are showing diminishing returns, and in a world where technology is becoming commoditized, trusted distribution is one of the few real moats left.

That's why I was so excited to sit down with Alex Lieberman, the co-founder of Morning Brew and now the CEO of Tenex. Alex is a master of distribution, but even he is capped on time. So, he re-engineered his entire content creation process to be AI-native, building a 'Content Machine' that not only scales his own output but also empowers his entire team to become creators. This isn't just about using AI to write; it's about using AI to build a system for authenticity at scale.

In this episode, Alex did something incredibly brave: he ran his Content Machine live. He showed us every step, from an AI 'Oracle' that finds content ideas to an 'Interview Panel' that pulls out his unique insights and a 'Writer's Council' that revises his posts. He argues that what we call 'AI slop' is often just a reflection of our own uninteresting ideas, and his system is designed to solve that at the source.

We're going to walk through the exact six-step workflow he uses, including the tools, the prompts, and the feedback loops that make it work. This is a masterclass in applying AI thoughtfully to a real-world business problem, and it's a blueprint for anyone who wants to build their brand without losing their voice.

Workflow 1: The AI-Powered Content Machine

Before Alex even touched an AI tool, he did the most important thing: he mapped out his existing content creation process. He literally drew out every step, from finding an idea to writing, editing, and distributing. This is a step I tell everyone to take. You can't improve a process until you understand it. Only then can you see where AI can be a driver, a co-pilot, or where a human needs to stay in full control.

After mapping the old way, he rebuilt the entire thing from the ground up, assuming no constraints. The result is this brilliant six-step system built primarily in Claude using a plugin architecture within Tenex's internal tools.

Step 1: The Oracle - Defeating the Blank Page

The biggest friction in content creation is the blank page. The Oracle solves this. It’s an idea generation engine that connects to all of Alex's systems of record.

• Tools: Connects via API to Slack, Notion, Gmail, Linear, and his social media feeds.
• Process: Every day, the Oracle scans the last seven days of information from two sources:

1. Internal Systems: It looks for 'content spikes' in his conversations, meeting notes, and project updates. It uses a scoring system that prioritizes ideas with strong stories, anecdotes, or contrarian points of view.
2. Internet Reader: It scans the latest posts from a pre-defined list of accounts Alex follows on X and LinkedIn, looking for opportunities to respond or add to a conversation.

• Result: Alex gets a daily ranked list of 15 content ideas. Even if he doesn't use the rest of the machine, this alone is a huge win. All unused ideas are automatically saved to a Notion database called 'The Vault' for future inspiration.

Step 2: The Interview Panel - Extracting Your Real Ideas

This is where Alex's system gets really smart. Instead of asking AI to write for him, he has it interview him. He believes the root of AI slop isn't bad AI writing, but bad human input.

"My take is that AI slop is hilariously people just pointing the finger at themselves and saying, 'I'm not intelligent enough.' The only time the content machine actually produces slop is more of an indictment of the person not sharing good enough ideas during the interview step."

• Process: Once Alex selects a content spike, the machine kicks off an 'Interview Panel' of six AI personas codified as skills: Tim Ferriss, Joe Rogan, Barbara Walters, Howard Stern, Michael Barbaro, and Larry King. One by one, they ask him probing questions about the topic. Alex simply talks through his answers, which are transcribed in real-time using Whisper Flow.
• Outcome: This process forces him to articulate his raw thoughts with specificity and examples. The entire transcript of this 'yap to text' session becomes the raw material for the draft. The AI isn't inventing sentences; it's structuring his own words.

Step 3: Drafting with a Codified Voice

To ensure the draft sounds like him, the Content Machine references a personal folder containing three critical markdown files:

1. `style-guide.md`: Defines who Alex is, his role at Tenex, and assets he wants to promote.
2. `voice-guide.md`: The secret sauce. This file is the result of an initial analysis of his all-time top-performing posts. It codifies his core DNA, hook formulas, content structures, and even specific language patterns like 'self-deprecating confidence.' It explicitly states his number one rule: 'Write like you're texting a friend.'
3. `content-lessons.md`: A log of all the mistakes the AI has made in the past. This file is central to the system's ability to learn.

When the drafting process begins, the AI uses the interview transcript as the source material and these three files as its instruction manual for tone, style, and structure.

Step 4: The Writer's Council - Automated Quality Control

Once a draft is ready, it goes to the 'Writer's Council' for an automated peer review. This is another panel of AI personas, this time modeled after great writers and thinkers.

• Process: Six personas, including David Perell, Sean Puri, Morgan Housel, and a character Alex created called the 'AI slop allergist,' read the draft. Each one scores it on a scale of 1 to 10 and provides feedback.
• The Revision Loop: If the aggregate score is below a 9 out of 10, the system automatically triggers a revision loop. It uses the council's feedback to improve the draft and resubmits it until it passes the quality threshold. This ensures a high standard is met even when Alex is in a hurry.

Step 5: The Lessons Loop - The Reinforcement System

This is how the machine gets smarter over time. After the Writer's Council is done, Alex gives the post a final review. He makes his own edits and then publishes.

• Process: After he's done, a job called the 'Lessons Loop' kicks off. It performs a diff to compare the AI's final draft with Alex's published version. It then tries to extract abstract, generalizable lessons from his changes.
• The Feedback: The system then asks Alex: 'Do you want these lessons to be added to the lessons file?' If he says yes, the feedback is permanently logged in the content-lessons.md file, ensuring the AI won't make the same mistake twice.

Step 6: Repurpose and Distribute

Once the anchor piece of content is finalized, the work isn't over. Alex can then ask the Content Machine to repurpose it, referencing the Gary Vaynerchuk content pyramid framework. He can simply say, 'repurpose this into three short tweets and two long LinkedIn posts,' and the machine will adapt the content for each format based on the styles defined in his voice guide.

Workflow 2: The Tenex Creator Cup

This AI-powered content machine isn't just for Alex. It's a tool he's giving to his entire company to fuel a brilliant employee advocacy program called the 'Tenex Creator Cup.' Alex believes his employees are his most underleveraged marketing channel and a key to winning the war for talent, especially as a bootstrapped company.

• The Game: It's a month-long challenge to get everyone at Tenex posting on LinkedIn and X. They use a dedicated Slack channel called 'Reply Guys' to share posts and cheer each other on.
• How it Works:
• Points System: You get 10 points for posting and 3 points for engaging with a teammate's post.
• Prizes: There's a $5,000 prize pool, with weekly games and editor's picks to keep everyone motivated, not just those with large existing followings.
• The Goal: Alex was clear that even if the Creator Cup generates zero business leads, it's a massive win. If it helps him hire just one top-tier engineer who discovered the company through an employee's post, the $5,000 investment pays for itself many times over compared to a recruiting agency fee.

This is such a powerful model for other companies. Instead of fearing employees building personal brands, Alex encourages it. He argues that the fastest way to lose a great employee is to prevent them from talking about the amazing work they do.

Personal AI Workflows

Beyond his professional life, Alex is using AI in some really thoughtful personal ways too.

• Custom Job Board: When his wife was job hunting, he built a personalized hiring board that emailed her 10 curated job openings every day, scored for relevance, and even included a button to autofill the initial applications.
• Children's Books: For his 11-month-old daughter, he's creating custom children's books with stories that are personalized to their family.

Final Thoughts

Alex's approach is a perfect example of how to build with AI, not just delegate to it. He has meticulously designed a system that uses AI to eliminate friction (the blank page), extract authenticity (the interview), enforce quality (the council), and learn over time (the lessons loop). He's not just automating content; he's building a scalable engine for his own expertise and empowering his team to do the same.

It’s a powerful reminder that the most effective AI applications are often the ones we design ourselves, mapping the technology to our unique workflows and goals. Now, what process can you map and re-engineer with AI today?

Thanks to Our Sponsors

• Firecrawl—Power AI agents with clean web data
• Customer.io—Build customer engagement campaigns from a single prompt

Episode Links

• Listen on Apple Podcasts, Spotify, and YouTube
• Find Alex on X and LinkedIn, and check out his company, Tenex.
• Find me, Claire Vo, on X and LinkedIn.
• Production and marketing by Penname. For inquiries about sponsoring the podcast, email jordan@penname.co.