Seedance 2...really?! Full AI UGC Workflow (Claude + Seedance)

Seedance 2...really?! Full AI UGC Workflow (Claude + Seedance)

How to Build a UGC Ad Testing Pipeline Using Claude NC and Cense 2 API

Introduction to the Tutorial

  • The speaker introduces a tutorial on building a User-Generated Content (UGC) ad testing pipeline using Claude NC and Cense 2 API, emphasizing its potential profitability.
  • Personal anecdote about transitioning from multiple supplements to AG1, highlighting the importance of gut health support with probiotics and digestive enzymes.

Importance of AG1

  • The speaker shares their experience with AG1 as their "one health secret," which includes 75 vitamins and minerals in one scoop.

Overview of Cense 2 API Pricing

  • Current pricing for Cense 2 API is discussed, noting that most users pay around $5 for a 15-second clip due to limited functionalities.
  • The speaker reveals how to achieve a significant discount on the API usage cost, reducing it from $0.03 to $0.18 per second while gaining full access features.

Tutorial Structure Clarification

  • This tutorial is not just a feature walkthrough; it's an in-depth breakdown aimed at those wanting to build workflows or pipelines rather than simply shipping AI ads.

Commitment to Knowledge Sharing

  • The speaker expresses their commitment to sharing knowledge without gatekeeping, contrasting this approach with larger companies that often restrict access behind complex features.

Required Tools for the Tutorial

  • Three essential tools are identified: Claude (app or code), Cense 2 API, and creativity along with an advertising-worthy product.

Setting Up Your Project Folder

  • Instructions begin on creating a project folder named "cense ab testing" where users will store their API key and documentation.

Creating an API Key

  • Steps are provided for accessing the Cense 2 dashboard, creating an API key labeled "cense test," and saving it in the project folder.

Documentation Setup

  • Users are guided on obtaining CES2 documentation optimized for Claude and pasting it into the same text file as the API key.

Pricing Comparison Insights

  • A comparison between Enhancer Cens2 pricing versus competitors highlights that Enhancer offers unlimited calls without queues at competitive rates.

Coding Interface Preferences

  • The speaker prefers using Claude's app over other coding environments like VS Code due to its user-friendly interface suitable for non-tech founders.

Testing Your Setup

  • Before proceeding with AB testing flow, users are instructed to drag their project folder into Claude and command it to set up the API.

Testing API Key and Image Generation

Setting Up the Test

  • The speaker discusses testing the API key by uploading an image to a temporary hosting service, tmpmp files.org, to generate a link for use in Cense.
  • The importance of having a URL for the image is emphasized, as Cense requires online links for processing various input types like images, audio, and video.

Running the Generation

  • The speaker prepares to run a generation request with specific parameters: a 5-second clip at 720p resolution and a 9x6 aspect ratio featuring a subject saying "hi."
  • A payload is defined as the settings sent to AI; it includes details like aspect ratio and resolution. This structure is based on documentation saved earlier.

Payload Verification

  • The payload includes essential elements such as type (image to video), model (multi-reference), prompt, duration, resolution, and aspect ratio.
  • A successful response (200 OK) indicates that the request was accepted and is being processed in the backend.

Tracking Generation Status

  • A request ID is provided with each payload submission, allowing tracking of the generation process until completion.
  • Users are advised to check their enhancer account under usage for real-time updates on requests made through Claude.

Creating UGC Ads Pipeline

Introduction to UGC Ads Pipeline

  • The speaker introduces an agentic system powered by Claude designed for generating User Generated Content (UGC) ads efficiently.
  • After three weeks of development, this pipeline will be shared freely with viewers.

Project Setup Instructions

  • Viewers are directed to download the UGC Ads Pipeline project from GitHub. They will unzip it into a folder named "Cense 2 UGC."

Onboarding Process

  • Upon running the project in Claude, an onboarding flow begins that checks setup status and provides welcome messages along with available models from the API.

Pricing Transparency

  • Detailed pricing information per resolution is displayed before any costs are incurred. For example, Cense 2 fast at 480p costs $0.073 per second.

Storing Sensitive Information

  • Users are instructed to save their API key in an ENV file within their project folder for security purposes; this prevents repeated manual entry during sessions.

Onboarding Process for AG1 Product

Step 1: Initial Setup

  • The onboarding process begins with saving the initial setup, ensuring no further adjustments are needed.

Step 2: Product Information

  • Claude requests detailed information about the product AG1, including its name, category (wellness), and description (boosts energy).
  • The user is unsure of key selling points and opts to let Claude suggest them instead. Price point set at $99, highlighting high quality with 75 vitamins.

Suggested Selling Points

  • Claude provides suggested selling points for AG1:
  • Contains 75 vitamins and minerals.
  • Features clean, high-quality ingredients.
  • Offers an all-day energy boost.
  • Supports gut health and immunity.

Step 3: Audience and Goals

  • The target audience identified as health-conscious adults; ad objective is conversion on platforms like TikTok and Instagram. Tone is energetic and aspirational. No discount code initially noted but later added as "25% off."

Brand Profile Creation

  • Claude saves the brand profile as a JSON file in the configuration folder, acting as a digital profile card that retains all relevant information about AG1 for future reference. This includes targeting details, selling points, discount codes, etc.

Control Center Overview

Accessing Control Center

  • The control center launches via a local HTTP link (port 8099), providing a simplified interface to manage ads without needing separate apps or subscriptions—just a project folder made easier to navigate.

Uploading Assets

  • Users can upload various assets such as product images, subject images, mood references, and audio clips directly into the control center for streamlined management of advertising materials.

Importance of Audio Clips

  • A six-second audio clip uploaded helps maintain consistent tone and pacing across generated ads featuring a specific spokesperson or brand face. This ensures uniformity in voice throughout variations created by the pipeline.

Asset Management Integration

  • Uploaded assets automatically create corresponding folders within the project directory (e.g., AG1 image in products folder). Deleting an asset from the control center also removes it from the local directory—showcasing full integration between both systems.

Brand JSON File Storage

  • A brand JSON file containing comprehensive details about AG1 resides in the assets folder for easy access during future interactions with Claude or when revisiting campaign strategies.

Ad Script Generation Process

Analyzing Assets for Context

  • The analysis of assets begins to add context for ad script generation, with Claude identifying three key assets related to the product and subject.
  • The product identified is AG1, while the subject is described as a young man with curly hair, a beard, and a smile. An audio clip has also been found.

AI Context Features

  • Accessing AI context provides detailed descriptions of visual elements and notes for both the product and subject; however, audio descriptions are not transcribed.
  • Confirmation of the brand profile includes essential details like product name, category, description, price, unique selling points, target audience, and goals.

Enhancing Brand Profile

  • The AI suggests adding more information such as CTAs or discount codes to enhance context; in this case, a discount code "save 25" will be added.
  • Updates to the JSON file include the new discount code and CTA link in bio for future generations.

Video Generation Setup

  • The process involves AB testing by specifying video quantity (eight), duration (15 seconds each), aspect ratio (9:16), and formats.
  • All configurations are set up automatically within the folder structure without manual intervention required from users.

Finalizing Ad Variations

  • Eight variations of ads are generated across different formats including podcast ads and UGC ads; cost estimates are provided for all variations.
  • A resolution adjustment is made from 480p to 720p before final submission to ensure quality standards are met.

Approval Process Before Submission

  • Emphasis on reviewing payload details before submission highlights that any mistakes could incur costs; explicit approval is necessary at every step.
  • Final confirmation shows an estimated cost of $22 for generating eight videos at the desired resolution.

Video Submission Process and Insights

Overview of Video Quality Options

  • The video submission process begins with selecting the quality; standard quality is set at 720p, while a cheaper fast mode is recommended for ads or social media assets, being approximately 30% less expensive.

Submitting Variants to Cense API

  • The system submits eight video variants to the Cense API, checking the API key and authorizing each version sequentially.
  • Each submitted variant receives a unique request ID, indicating that it is in queue for processing. Users can verify submissions by logging into Enhancer and matching request IDs.

Polling Mechanism Explained

  • The backend polling mechanism checks the status of video processing every two minutes without user intervention, ensuring automation in tracking completion.
  • Once a video completes processing, it automatically downloads to both the control center and designated folders.

Monitoring Video Completion

  • Users can manually refresh their dashboard to check on the status of videos; as videos complete, they are displayed along with prompts used for generation.
  • A total of eight videos will be available for preview once all are processed successfully.

Cost Efficiency and Additional Features

  • The cost per generated video is highlighted; users can generate multiple formats efficiently without needing to re-enter brand information or context.
  • New variants can be created using existing context from previous submissions, streamlining the process significantly.

Advanced Video Generation Capabilities

  • Users have flexibility in generating various styles (e.g., podcast format), leveraging saved brand details without repetitive input.
  • The system allows for creative freedom by enabling users to request specific styles like cinematic formats based on simple descriptions.

Addressing Potential Issues

  • While showcasing capabilities, there’s an acknowledgment that some generated videos may contain errors such as spelling mistakes or visual discrepancies.

This structured summary provides a comprehensive overview of the discussed topics within the transcript while maintaining clarity and accessibility through organized headings and bullet points linked directly to timestamps.

AI Output and Quality Control

Understanding AI Prompting and Testing

  • The speaker discusses the importance of recognizing that raw AI output may not be perfect on the first attempt, emphasizing the need for quality control in prompt generation.
  • It is recommended to start with a 780p fast setting at 5 seconds before committing to a full 15-second standard quality run, as fast settings are often sufficient and more cost-effective.
  • The process involves testing prompts first, iterating based on results, scaling up, and conducting A/B testing to avoid unnecessary credit expenditure.

Health and Productivity Enhancements

  • The speaker shares personal health improvements attributed to AG1 supplements, highlighting their composition of 75 vitamins, minerals, and probiotics that optimize bodily functions.
  • A humorous anecdote about friends encouraging the speaker to consume "the green stuff" illustrates the perceived benefits of AG1 in enhancing energy levels.

Project Management with Cense 2 API

File Management and Tracking

  • The control center mentioned serves as a visual interface over local project folders where all generated videos are stored alongside a matrix JSON file logging every payload sent during sessions.
  • Users can access their video history through the Enhancer API dashboard which tracks costs, request IDs, and allows for easy downloads tied to individual accounts.

Subscription Benefits

  • A subscription plan offers significant discounts (up to 75% off), providing premium access to features like unlimited concurrency for scaling ad creation processes effectively.

Future Content Creation Insights

Upcoming Tutorials and Community Engagement

  • The next video promises insights into creating cinematic commercials from single prompts while encouraging viewers to engage by liking and subscribing for better content visibility.
  • An invitation is extended to join a free community called Public AI where participants can collaborate on building real SaaS applications with support from experienced engineers.
Video description

Get up to 75% off Seedance 2 API — use code SAVE25 at checkout: https://app.enhancor.ai/api-dashboard or https://app.enhancor.ai/video-generator for in-app generations. Same price. Get the free UGC Ads Pipeline and start building today: https://github.com/sirioberati/Seedance-2.0-AI-UGC In this video I build a complete AI UGC ad testing pipeline using Claude and the Seedance 2 API from scratch. I show you how to access Seedance 2 at 45% off inside Enhancor.ai (0.155$ per second at 720p) with full human face upload support, 720p resolution, and unlimited concurrency — and how to stack code SAVE25 on top for up to 75% off total. Then I walk through a free GitHub project I spent three weeks building that lets you generate, AB test, and organize entire UGC ad campaigns directly inside Claude with no subscriptions, no UI clicking, and no complicated setup. You will learn⤵️ 1. How to access Seedance 2 API at 45% off inside Enhancor.ai and stack code SAVE25 for up to 75% off total 2. How to set up the Seedance 2 API inside Claude using just a text file and the documentation 3. What a payload is and why you need to review it before every generation 4. How to download and run the free UGC Ads Pipeline GitHub project inside Claude 5. How to build a brand profile that Claude saves and reuses across every single generation 6. How to upload product images, subject photos, and audio references through the control center 7. How to generate 8 ad variations across 4 formats including UGC, podcast, lifestyle, and green screen 8. How Claude polls the API automatically and downloads completed videos without you touching anything 9. How to iterate, add new formats, and keep generating without resetting context 10. How to verify every generation independently in your Enhancer API dashboard Tools and workflows in this tutorial⤵️ Seedance 2 inside Enhancor (45% off): https://app.enhancor.ai/video-generator Seedance 2 API Dashboard (use code SAVE25 for up to 75% off): https://app.enhancor.ai/api-dashboard Free UGC Pipeline on GitHub: https://github.com/sirioberati/Seedance-2.0-AI-UGC Free resources⤵️ PublicAI Community: https://www.skool.com/publicai Free Seedance 2 Prompt Engineering Guide: https://seedance-prompt-guide.sirioberati.com Seedance 2 Resource Hub: https://seedance.sirioberati.com AI UGC ads: https://ugc.sirioberati.com Wan Animate: https://docs.comfy.org/tutorials/video/wan/wan2_2 Camera Control Guide for Kling: https://camera.sirioberati.com Guide to Fix Fake AI Voice: https://influencer.sirioberati.com FREE Nano Banana Light Presets: https://presets.sirioberati.com Build landing pages with Claude: https://design.sirioberati.com Connect with me on: https://instagram.com/heysirio https://instagram.com/saysirio Chapters Theory 00:00 – 00:53 Intro: What We're Building and Why 00:53 – 01:36 The Real Problem With Seedance 2 API Pricing 01:36 – 02:34 Why This Tutorial Is Different 02:34 – 03:20 What You Need and How to Think About This Practice 03:20 – 04:32 Step 1: Creating Your Project Folder and API Key 04:32 – 04:57 Grabbing the Seedance 2 API Documentation for Claude 04:57 – 05:52 Pricing Breakdown: Enhancor vs Every Competitor 05:52 – 06:34 Setting Up the API Inside Claude 06:34 – 07:30 Hosting Your Image and Testing the API Live 07:30 – 08:44 Reviewing the Payload Before You Submit 08:44 – 10:38 Verifying Generations in Your Enhancer Dashboard 10:38 – 11:22 Step 2: Downloading the Free UGC Pipeline from GitHub 11:22 – 12:50 Running the Pipeline and the Onboarding Flow 12:50 – 14:22 Building Your Brand Profile with Claude 14:22 – 15:52 Audience, Goals, Discount Code and CTA Setup 15:52 – 17:12 Step 3: The Control Center Explained 17:12 – 19:22 Uploading Product, Subject, and Audio Assets 19:22 – 21:06 Analyzing Assets and Confirming Brand Profile 21:06 – 22:32 Finalizing the Brand Profile and Voice Matching 22:32 – 24:52 Generating 8 Ad Variations Across 4 Formats 24:52 – 26:16 Reviewing Payloads and Approving the $22 Generation 26:16 – 27:56 Submitting All 8 Variants to the API 27:56 – 29:14 Polling, Auto-Download, and Reviewing Results 29:14 – 31:06 Adding New Formats Without Resetting Context 31:06 – 32:52 Honest Take on Raw AI Output and How to Iterate 32:52 – 34:20 Where Your Files Live and the Matrix JSON Log 34:20 – 35:14 75% Off Seedance 2 API and How to Get It 35:14 – 36:14 Outro + PublicAI Pilot Program