Skip to main content
DanManREAL/ AI_LAB
Updated Nov 20, 2026Request a test

Built with AI4 min readBy Published Updated

MythREAL_FACTORY: lessons from my retired first video workspace

In this guide

The short version

  • Making videos alone means many separate steps, so I put them in one private workspace.
  • I built it in about thirty small, ordered steps, guided by a written plan.
  • Keep heavy, slow work separate from the everyday parts.
  • If a tool can imitate a voice, only use voices that are yours or that you have permission to use.
  • Save finished pieces so slow, costly work is never repeated.

Guide

4 min read
Illustration of a film clapperboard with a microphone and a paint palette

MythREAL_FACTORY is the workflow I use to make videos today. Before it, I built a private six-tab workspace with Claude Code. Making videos on my own meant a long chain of separate jobs, so I wanted one place to bring them together. This is the story of that earlier build: its plan, its order of work and a few decisions I would repeat.

I retired the earlier workspace on October 26, 2026. The story below stays as a historical record, rather than something you can use. To see the current workflow, watch the MythREAL_FACTORY walkthrough.

The problem: too many separate steps

If you make videos alone, the bottleneck is rarely one big thing. It is the chain: a script, a voiceover, thumbnails, visual polish and the files you need to publish. Each step takes time, and each can stall the others.

I was jumping between disconnected tools for all of it. What I wanted was one place where a project could move from idea to finished video without leaving the room.

The idea: one workspace, one tab per step

The result was a private workspace with six tabs, one for each stage of the job. It was not one AI button with six labels. It was a set of tools that carried a project forward.

TabWhat it did for me
ScriptDrafted the story that everything else built on.
StoryboardPlanned the pacing and the order of scenes.
VoiceMade narration quickly.
ThumbnailHelped me design the picture that gets the click.
Hero matching and libraryFound the game-specific pictures and assets I needed.
Editor and exportPut the pieces together and prepared the final delivery.

It sat behind my private login, so it was never something you could visit. It existed so my own videos took less effort.

Building in small steps, not one giant request

I did not ask the assistant for "the whole workspace" in one go. I built it in about thirty small, ordered steps. Each step had a clear boundary, so I knew what it should touch and what it should leave alone.

A written plan held it together. I kept the planning notes, feature notes and rules outside the assistant, on the page. Each session then began from the same plan and did not invent a new direction.

  1. Write down the plan and what the finished thing should do.
  2. Cut it into small steps, each with a clear boundary.
  3. Do one step at a time and check it before moving on.
  4. Update the notes so the next step starts from what is true now.

Big builds do not usually fail because the assistant cannot write code. They fail when the sequence loses the plan. Later requests then spend their effort undoing earlier ones.

Try thisA long build survives on order, not heroics. If a request has no clear boundary, split it.

Keeping heavy work out of the way

Some jobs are simply heavier than the rest. Turning speech into text, cutting a person out of a background and heavier video effects all take real computing power. I gave them their own separate lane.

The idea is to keep the everyday parts quick, put the expensive work to one side and draw a clear line between the two. That also makes problems easier to trace. The workspace also stayed separate from my public pages.

The voice tab, used with care

The voice tab turned a finished script into spoken narration. Tools like this can sound very close to a real person, which is why they need a firm rule.

If you ever build something that copies or imitates a voice, set that rule first: only voices that are yours or that you have clear permission to use. Write it down where your assistant will read it.

Saving finished pieces so slow work is not redone

Making narration is slow and costs money each time. A demo can afford to redo it every time. A tool you use repeatedly cannot.

So the workspace saved what it produced. When I needed a line of narration I had already made, it reused the saved one instead of making it again. Finished narration, repeatable assets and files ready to export built up over time.

That habit is what separates a one-off demo from something you can rely on.

What I learned

  • Order beats big effort: many small, checked steps outperform one giant request.
  • Keep the plan on the page, not in the assistant's memory.
  • Separate the slow, heavy work from the everyday parts.
  • A big private tool can live alongside a public site if the two stay clearly separate.
  • Save finished work so you never pay for the same thing twice.

The same discipline shows up in my other builds. I describe the habits behind it in Keep your AI coding assistant on track, and you can see how it worked on a data project in How I built a 204-hero fan database with Claude Code. If you want to try building something yourself, start with Claude Code for gaming creators.

Common questions

What was the MythREAL workspace?

A private workspace with six tabs, one for each stage of making a video: script, storyboard, voice, thumbnail, hero matching and library, and editor and export. It sat behind my private login.

Is the MythREAL workspace still in use?

No. I retired it on October 26, 2026. My episodes are now made by MythREAL_FACTORY, a local AI video pipeline.

How do you build a big tool with an AI coding assistant?

In small, ordered steps, not one giant request. I built the workspace in about thirty steps, each with a clear boundary, guided by a written plan kept outside the assistant.

Why save finished pieces like narration?

Making narration is slow and costs money each time. The workspace saved what it produced, so a line I had already made was reused instead of made again.

Next step

See what these tools actually did

The guides explain how the tools work. The lab shows what happened when I used them on real work: verdicts, costs, failures and the receipts behind each one.

Keep reading