Descriptions:
Alex Lieberman, founder of 10x (an applied AI consultancy serving mid-market and enterprise clients), joins Claire Vale on the How I AI podcast to walk through his company’s AI-native content creation system. Built to solve two problems — maximizing output for a seasoned content creator and lowering the barrier for employees to publish — the system demonstrates a structured, multi-step approach to AI-assisted writing that avoids generic output.
The workflow begins with a component called the Oracle, which scans the past seven days of information across Lieberman’s channels and scores content ideas using a structured ranking system, surfacing roughly 15 “content spikes” per cycle. An interview step then captures the creator’s raw thinking as voice or text, which becomes the primary source material. A voice guide — built by analyzing Lieberman’s top-performing posts on X and LinkedIn — codifies tone, hook formulas, sentence patterns, and topic categories so the AI writes in the creator’s authentic voice rather than generic AI prose.
A reinforcing feedback loop called the “lessons loop” captures edits made to drafted content, extracts abstractable writing lessons, and appends them to a persistent lessons file checked before every future draft. Lieberman’s core argument is direct: AI content slop is primarily a failure of input quality from the human, not a model capability problem. The session includes a live demonstration building a content piece in real time, giving viewers a concrete picture of how the Oracle, voice guide, and lessons loop interact in practice.
📺 Source: How I AI · Published July 20, 2026
🏷️ Format: Workflow Case Study
