In the last 12 months, my content output has tripled while my actual time spent writing and editing has dropped by half. This isn't because I'm copy-pasting generic ChatGPT outputs. It's because I've built a "Human-in-the-Loop" (HITL) AI workflow that uses tools like Claude, Notion AI, and Descript to handle the heavy lifting, leaving me to focus purely on strategy, voice, and experience.

Back when I was illustrating for Disney and Marvel, the idea of an "automated workflow" meant setting up a few clunky Photoshop actions. Today, running my own AI consultancy while managing my charity, Become Inspired, across 12 different countries, efficiency isn't just a nice-to-have—it's survival. I know firsthand that generating volume without value is a surefire way to kill a brand. In 2026, the trend has definitively shifted from mass-producing cheap AI content to building sophisticated "content waterfalls"—taking one high-value human asset and using AI to perfectly repurpose it. Here is the exact, step-by-step workflow I use to do this today.

Step 1: Automated Research & Ideation with Perplexity and Claude

Every piece of content starts with data, not guessing. Before I type a single word, I use AI to validate the idea.

  1. Search Intent Mapping: I use Perplexity to search for my target topic. Unlike standard search engines, Perplexity gives me a synthesis of the top 10 articles, highlighting exactly what questions people are asking right now.
  2. The "Devil's Advocate" Prompt: I feed my initial idea into Claude 3.5 Sonnet and prompt it: "Play the role of a cynical senior editor. Tell me why this article idea is boring, what it's missing, and what unique angle I need to take to make it stand out in 2026."

This instantly forces me to elevate my content above the generic AI baseline.

(For more on how I use Perplexity, read my deep dive: How I Use Perplexity AI for Market Research.)

Step 2: The "Brain Dump" First Draft (Using Voice)

The biggest bottleneck in content creation is staring at a blank page. My solution? I don't type my first drafts anymore; I speak them.

Using ChatGPT's Voice Mode or an integrated transcription tool, I pace around my home office and simply explain the concept out loud for 10 minutes. I talk about my real-world experiences, the mistakes I made, and the specific outcomes.

I then feed that raw, rambling transcript into Claude with my custom brand voice prompt: "Reorganize this transcript into a coherent blog post outline. Preserve all of my personal anecdotes, specific numbers, and my direct, slightly informal tone. Do not add any filler words."

Step 3: Refinement in Notion AI

Once the draft is structured, I move it into my central workspace: Notion.

Notion AI acts as my embedded assistant. Instead of switching tabs, I use inline commands to tighten the prose. My most frequently used commands are:

  • /ai make shorter (applied to overly long paragraphs)
  • /ai simplify language (to remove any lingering "AI-speak" or corporate jargon)

This is where the "Human-in-the-Loop" aspect is critical. I spend 80% of my writing time here, manually editing, adding nuance, injecting humor, and ensuring the final piece sounds undeniably like me.

Step 4: The Content Waterfall (Descript & Opus Clip)

If the content is a video or podcast, the workflow shifts to the "waterfall" method—repurposing one core asset into dozens of micro-assets.

  1. Main Edit: I load the raw video into Descript. It transcribes the audio, and I edit the video by deleting text (which automatically cuts the video). Descript’s Studio Sound feature instantly masters the audio.
  2. Micro-Content: I export the finished video and run it through Opus Clip or Runway. These tools automatically identify the most engaging 30-60 second segments, crop them for vertical formats (TikTok, Reels, Shorts), and add dynamic captions.

What used to take an editor two days now takes the AI 15 minutes.

The Verdict: Quality Over Quantity

The trap most creators fall into is using AI to replace themselves. My workflow uses AI to replace the friction, not the author.

What worked? Using AI for research, structuring transcripts, and video repurposing. What didn't work? Asking AI to write the final draft from a blank prompt. The internet is already flooded with generic AI content; your unique, lived experience is the only differentiator that matters in 2026.

cluster: "Content Strategy"

Frequently Asked Questions

What is a Human-in-the-Loop (HITL) AI workflow?
A HITL workflow means using AI for repetitive or mechanical tasks (like transcription, outlining, or formatting) while a human remains actively involved in the editing, strategy, and final quality control.

Which AI is best for writing blog posts in 2026?
Rather than relying on one tool, the best approach is a stack: Perplexity for research, Claude for structuring raw thoughts and maintaining brand voice, and an integrated tool like Notion AI for final line editing.

What is a content waterfall in marketing?
A content waterfall is the process of taking one large, high-value asset (like a podcast or long article) and using AI to automatically break it down into smaller pieces (tweets, vertical videos, LinkedIn posts) for distribution.

cluster: "Content Strategy"

Author Bio

Simon Bullows is an AI & Creative Consultant helping solo-consultants, charities, and non-technical teams leverage artificial intelligence. With a background designing for Marvel and Disney, and as the founder of the Become Inspired charity, Simon writes about the intersection of creative workflows, remote work, and generative AI. Work with Simon.