Before you can sell anything else, you have to be able to tell your own story consistently. Most people skip that step and wonder why their brand feels different every time they post. Gasp House is a working AI ad agency — ten departments, a real creative-review process, an agency that actually rejects its own bad ideas before they ship — built to run inside your own AI tool. Point it at a product and it'll build you a brief, an ad, a media plan, and a signed creative package. Point it at yourself first. Feed it your story instead of a product launch, and it'll do the same job on you: force a clear point of view, build you an actual brand reference instead of a vibe, and hold every future post to the standard of that reference — instead of relaunching your personality from scratch every time you sit down to write something. It's free. Not a stripped demo — the whole working system. This is the first piece of something bigger: a full department-by-department AI film studio, where Gasp House is one working department among several. That's still being built. This part's ready now, and it's yours. [Download the starter kit] HOW TO USE IT You'll need Claude Code or ChatGPT Codex already set up — both read the folder's instructions automatically, no configuration needed beyond having the tool installed. (No coding agent yet? Full setup steps, and a manual workaround for plain chat tools, are in QUICK_START.md inside the folder.) Unzip the folder, then open it with your AI tool (cd into it and run claude, or open it in Codex). Check it loaded correctly. Say: "Read CLAUDE.md and introduce yourself — who are you, and what happens between me giving you an idea and getting a finished deliverable?" If it answers specifically instead of guessing, you're set. Start with yourself, not a client. Say: "Open PERSONAL_BRAND_MODE.md and follow it. I'm running myself through this as the first project." Then answer honestly when it asks who you are, who you're trying to reach, and what you actually want right now.

Okay, I saw a post. People were loading in their ChatGPT generated storyboards into Seedance 2.0 and it was animating their sequences. So I tried it. UMMMM This is amazing. Storyboards I used attached too. Here is the first pass of the animation. Incredible!

Thank you!

For professionals utilizing artificial intelligence to manage creative workflows, unstructured data is a critical liability. Advanced tools like Codex and Claude Code are exceptionally capable, but they cannot inherently deduce the context of a 50GB raw video file, a high-resolution design asset, or a complex project directory based solely on a filename. When files are deposited into a drive without structural context, it forces the AI to operate on assumptions, which leads to inaccurate outputs and wasted time. The solution is an architecture known in AI engineering as the LLM Wiki or Open Knowledge Format. By standardizing a system of lightweight, structured text files adjacent to your creative assets, you provide AI agents with a definitive, machine-readable roadmap of your entire archive. Here is the exact framework to implement this standard across your local and cloud drives. 1. The Directory Blueprint: _index.md Every primary folder requires an _index.md file. This document serves as the executive summary for the directory, establishing the project's core objectives, visual psychology intentions, and structural hierarchy. For example, a directory dedicated to an analog short film project would contain an _index.md that dictates the project timeline, the structural breakdown of the subfolders, and the foundational aesthetic principles the AI must adhere to when referencing that project. 2. The Asset Blueprint: The "Sidecar" Non-text files require immediate context. You achieve this by creating a Markdown file with the exact same name as the media asset, establishing a "sidecar" relationship. If you have a file named Scene_01_16mm_Scan.mov, you create a corresponding Scene_01_16mm_Scan.md in the exact same location. This sidecar file contains the specific metadata: shot lists, focal lengths, lighting diagrams, and narrative context. When the AI scans the directory, it reads the text sidecar to fully understand the contents and intent of the adjacent media file.