How to Build an AI Content Creation Workflow

TL;DR

Map the full content workflow before adding AI, then use AI to find ideas, conduct interviews, preserve the creator’s voice, review drafts, and learn from feedback. Generic output usually reflects weak source material, so the system should extract genuine expertise and lived experience before drafting rather than expecting a model to invent compelling ideas.

Transcript

Creating content has afforded me so many opportunities, but I am capped on the amount of time that I can spend creating content every day. And so, I've been basically thinking about how can I re-engineer my content process to be AI native or AI assisted. The other problem I've been trying to solve is how do I make it as easy as humanly [music] poss... Read More

Key Insights

• Trusted distribution is a durable business advantage because technology is becoming more commoditized and traditional business moats are becoming fewer. Lieberman therefore treats content as more than promotion, aiming to build a media company on top of Tenex and strengthen its direct relationship with audiences.
• Employee expertise is an underused marketing channel because team members already possess knowledge about their work, customers, and industry. The Content Machine is designed to reduce the effort required for employees to share that expertise while continuing to perform their full-time roles.
• Workflow mapping is the starting point for effective AI adoption because content creation contains distinct stages with different needs. The process includes inspiration, optional research, raw idea capture, format selection, drafting, editing, publishing, repurposing, and distribution across platforms where audiences consume content.
• The blank page is a major source of friction because creators often struggle to find ideas that feel novel rather than recycled. The Oracle addresses this problem by scanning the previous seven days of information, scoring promising content spikes, and producing a shortlist of 15 ideas.
• Strong AI-assisted writing begins with strong human input because the interview stage supplies the ideas, experiences, and opinions that make a post distinctive. Lieberman argues that generic output often indicates that the person did not contribute sufficiently substantive material before the system began drafting.
• Voice and style files are tools for preserving a creator’s recognizable register during drafting. By codifying how the person communicates, the system can work from explicit guidance instead of defaulting entirely to patterns derived from the broader body of writing on which the model was trained.
• AI can improve some stages of content creation without improving every stage equally. Lieberman accepts that exceptional writers may lower their drafting or editing ceiling with AI, but argues that assistance with other steps, including discovery and process support, can still raise overall output.
• Repurposing multiplies the value of an anchor piece because one substantial work can become smaller pieces for multiple social platforms. This content-pyramid approach separates the creation of the central idea from its later adaptation and distribution wherever relevant audiences already consume information.

Questions & Answers

Q: How do you build an AI content creation workflow?

Start by mapping the existing process from beginning to end. The workflow described includes finding inspiration, conducting research when lived experience is insufficient, capturing raw thoughts, choosing an anchor format, drafting, reviewing, revising, publishing, and converting the anchor piece into smaller posts. AI can then be assigned to specific bottlenecks, including idea discovery, interviewing, voice guidance, editorial evaluation, and learning from feedback.

Q: How can AI-generated posts avoid sounding generic?

AI-generated posts become less generic when they are grounded in substantive human ideas, lived experience, and clearly documented stylistic preferences. The interview stage should draw out specific opinions and knowledge before drafting begins, while voice and style files guide the model toward the creator’s register. An editorial review process and a feedback loop can then score, revise, and improve later drafts.

Q: What does the Oracle do in the Content Machine?

The Oracle handles idea discovery by scanning information from the previous seven days across the channels available to the creator. It applies a scoring system that defines and ranks promising content spikes, then returns a shortlist of 15 ideas. Its purpose is to remove blank-page friction and give the creator timely starting points rather than requiring inspiration to appear on demand.

Q: Why is the interview step important for AI writing?

The interview step is important because it supplies the original thinking that the drafting system cannot reliably invent for the creator. It draws out expertise, personal experience, opinions, and details that distinguish a post from average internet writing. Lieberman argues that apparent AI slop often reflects inadequate ideas shared during this stage, rather than a failure of prose generation alone.

Q: How should a company introduce AI into content creation?

A company should first document how content is currently produced instead of treating creation as one undifferentiated task. It can then identify where employees lose time or momentum, such as finding ideas, organizing thoughts, drafting, editing, or repurposing. AI assistance should target those stages while preserving employee expertise as the source material and maintaining review before publication.

Q: Why should employees become content creators?

Employees can become valuable creators because their work gives them relevant expertise and experiences that can support a company’s distribution. Lieberman sees employees as an underused marketing channel, especially when conventional channels produce diminishing returns. Since employees also have full-time responsibilities, the practical challenge is reducing creation friction enough that sharing their knowledge becomes manageable alongside their normal jobs.

Q: Can AI lower the quality ceiling for strong writers?

AI may lower the ceiling when an exceptional writer relies on it directly for drafting or editing because a model trained on a broad body of internet writing tends toward average patterns. However, content creation has many other stages. AI can still raise overall effectiveness by supporting idea discovery, research, organization, repurposing, and workflow execution without replacing the writer’s strongest creative abilities.

Q: Why is distribution becoming an important business moat?

Distribution becomes more important when technology is increasingly commoditized and fewer traditional business moats remain. Lieberman argues that building a good product does not ensure that customers will discover it. A trusted audience gives a company a direct way to communicate, and a media operation built around the business can strengthen that advantage through founders, dedicated creators, newsletters, and participating employees.

Summary & Key Takeaways

• Alex Lieberman rebuilt his content process because content had created many opportunities, yet he could devote only 25% of his time to it. He also wanted Tenex employees to create while keeping their full-time roles, making an AI-assisted system a practical way to reduce friction and expand trusted distribution.

• The conventional workflow begins with finding inspiration, researching when personal experience is insufficient, recording raw thoughts, selecting an anchor format, drafting, editing, publishing, and repurposing the anchor piece into smaller posts. Mapping these stages first reveals where AI can remove bottlenecks without automatically controlling every creative decision.

• The Content Machine uses an Oracle to surface recent content spikes, an interview panel to extract substantive ideas, voice and style files to guide drafting, an editorial council to score and revise posts, and a lessons loop to retain feedback. Its quality ultimately depends on the creator contributing specific, worthwhile thinking.