Agentic AI for Video Creation

Build a governed, end-to-end agentic video-production workflow—from evidence-backed research and generative content to human approval, SocialPost/YouTube release, and performance feedback.

Course detail	Information
Course code	TGS-2023036088
Programme	WSQ
TSC	MED-MPN-4005-1.1 — Video Editing-4
Duration	2 days / 16 training hours, plus 2 hours of assessment
Registration	View course details and register
Funding	Up to 70% course-fee support may apply. Eligibility and terms apply; verify the current course page before enrolment.
Courseware version	v2.0 — 29 August 2026

About the course

This technical course treats video creation as a controlled production system rather than a chain of unverified prompts. Learners use n8n to preserve evidence and state across research, scripting, storyboard generation, generative-media requests, assembly, quality review, human approval, dry-run publishing, and analytics.

The lab scenario uses synthetic data and safe defaults. External publishing remains disabled or in dry-run/private mode unless a trainer authorises a controlled demonstration.

Learning outcomes

By the end of the course, learners can:

1. Develop editing strategies and work plans that translate creative intent into a controlled agentic video-production contract.
2. Assess AI-generated and edited footage against storytelling, technical, brand, provenance, and platform-compliance evidence.
3. Develop remedial actions, approval controls, and n8n automations that support safe adoption of emerging video technologies.

Topics covered

1. Creative strategy and end-to-end production

• Production contracts, finish conditions, run state, and idempotency
• Evidence research, source scoring, claim provenance, and bounded prompts
• Timed scripts, storyboards, shot grammar, continuity, routing, cost, and latency controls
• Coordinator, specialist-agent, deterministic-validator, and human-owner boundaries

2. AI-assisted editing, storytelling, and quality assurance

• Generative video, image-to-video, avatar, voice, music, and caption contracts
• Asset manifests, rights status, checksums, and immutable versions
• FFmpeg assembly, WebVTT captions, brand overlays, and ffprobe technical checks
• Narrative QA, timecoded repair, and payload-hash human approval

3. Workflow optimisation, compliance, and publishing

• n8n triggers, expressions, sub-workflows, retries, timeouts, and secret isolation
• SocialPost multipart upload contract and YouTube private-release controls
• Synthetic-media disclosure, rights, privacy, observability, and publication reconciliation
• Retention analysis, controlled experiments, and human HOLD, ITERATE, or SCALE decisions

Labs

Complete the labs in order. Every folder contains a detailed README, an importable workflow.json, a styled synthetic mock-data.xlsx, an instruction-and-prompts PDF, starter notes, expected output, and an evidence checklist.

1. Create the Video Production Contract and Research Backlog
2. Run Evidence Research and Build the Claim Register
3. Generate the Script, Storyboard and Video Prompts
4. Build the Generative Video, Voice and Music Asset Pack
5. Assemble and Probe the Captioned Vertical Video
6. Run Quality Review and Human-in-the-Loop Approval
7. Orchestrate the End-to-End SocialPost and YouTube Release
8. Analyse Performance and Build the Scaling Control Plan

The connected path is:

contract → research → script/storyboard → generative assets → assembly
         → QA and human approval → SocialPost/YouTube dry-run → analytics

Public package

• Learner Guide in Markdown
• Eight self-contained lab folders
• Sanitised n8n workflow exports with no live credential values
• Synthetic Excel data, bounded prompt PDFs, solution structures, and evidence checklists

The editable slide deck, rendered course documents, and learner assessment papers are distributed through the authorised course Drive/LMS channels.

Distribution boundary

This public repository intentionally excludes:

• assessment answer keys and marking guides;
• .env files, credentials, tokens, or private connection details;
• source/reference material with restricted distribution;
• build tooling, QA renders, archives, and trainer-private resources.

Do not add live API keys to n8n JSON, prompts, workbooks, screenshots, issues, or pull requests. Configure approved credentials inside the target n8n environment.

Provider

Developed by Tertiary Infotech Academy Pte Ltd (UEN 201200696W).

• Course registration and current funding information
• SocialPost platform