The meeting provides a comprehensive, step-by-step walkthrough of setting up an AI-powered automation workflow to convert long-form podcast or YouTube video content into multiple viral social clips (optimized for Instagram Reels, TikTok, and YouTube Shorts) with automatic face detection and captioning.
The workflow leverages Airtable, Make (Integromat), open-source NCA Toolkit APIs, cloud hosting (Digital Ocean/Google Cloud), and AI models (ChatGPT/Claude) for automation, transcription, clip identification, editing, and captioning.
Key decisions include using Digital Ocean for server reliability, segmenting automations for error tolerance, and using deterministic tools over AI where precision is required.
Attendees: Not specified, but the video creator acts as facilitator and demonstrator.
Action Items
As soon as possible – All users: Duplicate the free Airtable template from the provided link and set up the required views and automations as demonstrated.
As soon as possible – All users: Set up and configure a Digital Ocean (or Google Cloud) instance; install and test the NCA Toolkit with correct environment variables and API keys.
As soon as possible – All users: Validate cloud storage (Digital Ocean Spaces) setup with access keys for storing generated media and transcription files.
After toolkit installation – All users: Run Postman tests to authenticate and verify the NCA Toolkit API is online using the provided API key and endpoints.
After verification – All users: Step through each Make automation (transcription, clip-identification, cutting, cropping, and captioning) to ensure end-to-end processing works correctly, adjusting filters/limits as needed for testing.
Ongoing – All users: Troubleshoot any errors in data formatting, API requests, or automations per the video’s error-handling/debugging advice.
System Demo/Overview & Initial Setup
The automation ("Content Clip Magic") ingests a video link, desired output dimensions, and triggers a sequence of processes to split long-form video into short, engaging clips ready for social media.
The workflow uses Airtable for data management, Make for process automation, the NCA Toolkit for free/open-source media processing, and AI models for transcript analysis and clip selection.
Users are guided through duplicating the Airtable base/template, preparing test rows, and configuring required automations.
Digital Ocean is chosen for hosting the NCA Toolkit due to better handling of long-running media jobs (compared to Google Cloud timeouts); cloud storage buckets are used to store intermediate and final files.
NCA Toolkit API Installation, Testing & Integration
Users must set up a containerized app using Docker Hub on Digital Ocean (settings and environment variable instructions provided in detail).
Cloud storage (Spaces) and access/secret keys are configured for file transfer and storage.
NCA Toolkit is tested using Postman collections to confirm valid authentication and file processing endpoints; successful response codes indicate readiness.
If issues arise, users are advised to meticulously check environment variables, API keys, and setup steps.
Airtable & Make Automations — Transcription, Clip Identification & Processing
Airtable is used to hold video metadata, transcripts, SRT files (timestamped captions), and resulting clips.
The first Make automation scans for new videos, sends them to the NCA Toolkit for transcription, and saves back the text/SRT into Airtable.
A second scenario uses AI (Claude or ChatGPT) to analyze the full transcript and propose several compelling segments (clips), with a fallback to deterministic parsing for better reliability in SRT start/end matching.
Additional modules parse and match transcript excerpts to their precise SRT segments for accurate clip cutting.
Clips are inserted back to Airtable, linked to their source video for downstream processing.
Cutting, Cropping & Finalizing Clips
Subsequent automations extract each identified clip from the long-form video using the calculated start times and durations.
OpenAI Vision (or similar) analyzes the clip thumbnail to determine the facial position (x/y coordinates) and computes optimal cropping rectangles for vertical video formatting.
The NCA Toolkit handles the actual cropping/scaling, outputs new video files, and updates Airtable with the processed URLs.
Final automation runs the captioning service on each cropped clip, producing social-ready, auto-captioned short videos.
Error Handling, Testing, and Scaling
Emphasis is placed on iterative testing: users are advised to run automations with test limits (process one record at a time) before scaling up.
Failure cases (e.g. AI errors, SRT/JSON formatting issues) are surfaced, with clear debugging procedures and a note that the system is robust to individual clip errors due to its modular, repeatable design.
Community & Template Resources
Users are encouraged to join the No Code Architects Community for access to ready-made blueprints, prompt templates, and technical support.
Decisions
Use Digital Ocean for NCA Toolkit hosting — Chosen over Google Cloud to avoid timeouts and for cost-effective, reliable processing of large video files.
Mix of AI and deterministic tools — AI (Claude/ChatGPT) is used for semantic clip selection; deterministic parsing is used for precise SRT segment matching, optimizing reliability and output quality.
Segment automation for error tolerance — Each automation step is independent, enabling easy debugging and robustness to single-clip failures.
Open Questions / Follow-Ups
Will future updates be needed to accommodate additional social media formats or platforms?
Are there best practices for optimizing server costs/performance as the number of processed videos increases?
How should the workflow be adapted if overlapping speaker faces or complex layouts are present in the video?
What routine checks or maintenance are recommended to ensure the automations run smoothly as Airtable or NCA Toolkit versions update?