Quick Overview
This video is a step-by-step tutorial presented by Nate Herk on automating video editing workflows with AI. Herk demonstrates how to configure the Codex desktop app with the Hyperframes toolset to create 3D motion graphics, cuts, and subtitles using plain language prompts. The walkthrough was produced to provide a practical build-along for creating edited video content without manual timeline editing.
Key Points
- 1.Codex paired with Hyperframes enables natural language prompting to automate video editing, cuts, transitions, and 3D motion graphics.
- 2.The core editing process operates on a five-step loop: transcribing audio, cutting pauses and mistakes, planning scene beats, rendering motion design via Hyperframes, and verifying output bounds and timing.
- 3.Speech-to-text transcription can be handled locally using Whisper or through an external API key from ElevenLabs stored in an environment configuration file.
- 4.Custom prompt instructions and operational rules are stored in an AGENTS.md file to standardize future video builds and workflows.
- 5.Users can make visual adjustments directly inside the Hyperframes browser interface or iterate on versions using structured goal prompts in Codex.
Summary
The video opens with Nate Herk demonstrating finished video samples edited entirely through natural language commands, highlighting automated cuts, subtitles, dynamic camera zooms, and 3D motion graphic overlays.
To build these assets, Herk sets up a workspace in the Codex desktop application and connects it to the Hyperframes repository. He explains that video generation follows a structured five-stage framework.
- 1.Transcribe. The AI transcribes the spoken audio to the exact millisecond using local tools like Whisper or external services like ElevenLabs.
- 2.Cut. The model identifies and removes speech mistakes, stuttering, and silence gaps lasting longer than half a second to maintain a fast-paced delivery.
- 3.Plan the beats. The system reviews the transcription to understand narrative intent, dividing the timeline into distinct visual scenes or beats aligned with the spoken script.
- 4.Use skills/hyper. Codex executes specialized skills and calls Hyperframes to generate animated HTML elements, 3D cards, icons, and text overlays synced to the audio.
- 5.Verify. The agent reviews the generated video, checking screenshots and playback against safety bounds and timing rules before finalizing the render.
Herk demonstrates the setup process by creating a local folder named Hyperframes demo in Codex and pasting the official repository URL. To configure high-speed transcription, he creates a dot env file, generates an ElevenLabs API key with speech-to-text access, and stores it in the project workspace. He also establishes custom guidelines in the AGENTS.md file so Codex remembers project preferences.
Next, Herk provides a raw unedited clip and submits a detailed goal prompt. He specifies exact visual parameters, including Apple-style liquid glass motion cards on the left, 3D rotating graphics on the right, word-synced subtitles in the lower third, a background overlay at 50 percent opacity, and a rounded face camera crop with drop shadows. Codex processes the request and returns a complete 28-second edit cut down from over a minute of raw video.
After reviewing the initial draft, Herk provides revision feedback to enlarge the text, separate overlapping 3D card animations, and add a dynamic full-screen graphic hook in the opening seconds. Codex updates the project files and renders a second version. Herk shows how users can fine-tune positioning within the Hyperframes local browser canvas or save high-performing prompt templates as reusable skills for future automated edits.
Overview of AI Video Editing with Codex and Hyperframes
Nate Herk demonstrates how natural language instructions can drive automated video editing workflows inside the Codex desktop app. By connecting Codex with the Hyperframes repository, creators can generate 3D motion graphics, Apple-style liquid glass cards, subtitles, background overlays, and multi-track cuts without traditional timeline editing software.
The Five-Step AI Editing Workflow
The editing pipeline follows a structured five-step loop: transcribing source footage, removing dead air and filler words, planning visual scene beats according to speech context, rendering HTML animations with Hyperframes, and running automated verification passes to ensure graphics and audio remain aligned.
Environment Setup and API Integration
To configure a new project in Codex, users clone the Hyperframes repository into a local directory and create a environment configuration file containing an ElevenLabs API key for accelerated transcription. Project-wide behavioral instructions and default dependencies are documented inside the AGENTS.md file to maintain consistency across editing sessions.
Prompting, Reviewing, and Iterative Refinement
Using Codex goal prompts, specific directions are given regarding camera layout, subtitle positioning, background opacity, and 3D animation timing. Once a draft renders, creators can either adjust layout elements directly inside the Hyperframes preview studio or submit corrective natural language prompts to produce updated versions.
The Bottom Line
The guide demonstrates how developers and creators can build an automated video post-production pipeline by combining Codex, Hyperframes, and transcription APIs. It shows that detailed prompting and prompt iteration yield production-ready short-form and long-form video edits with synchronized 3D graphics. While manual adjustments remain possible in the local studio interface, the video leaves broader multi-speaker editing and complex sound design workflows for future exploration.
FAQ
What is Codex and how does it integrate with Hyperframes for video editing?
Codex is an AI desktop environment that uses language models to execute code and terminal tasks, while Hyperframes is a repository that renders HTML-based animations, motion graphics, and video compositions driven by Codex prompts.
How does the five-step video editing workflow operate in Codex?
The workflow transcribes speech, cuts pauses and mistakes, plans scene beats, generates motion graphics with Hyperframes, and executes a verification loop to check layout and timing alignment.
What transcription services can be configured within the Codex Hyperframes setup?
Transcription can run locally using the Whisper engine or via an external ElevenLabs API key added to the project environment file.
What is the purpose of the AGENTS.md file in the editing project workspace?
The AGENTS.md file contains project documentation, workspace rules, and instructions that inform the AI agent how to handle tasks like default transcription tools and rendering procedures.
How can creators refine the video layout generated by Hyperframes?
Creators can refine videos either by submitting follow-up natural language goal prompts in Codex or by directly adjusting element properties and positions inside the Hyperframes browser interface.
Worth watching for
Video creators, developers, and digital marketers seeking to automate video post-production and motion graphics generation using AI agent workflows.
- video-editing
- codex
- hyperframes
- ai-agents
- motion-graphics
- elevenlabs