AGENC-IA Framework
How I use a team of custom-built AI agents as my second brain to organize marketing, write high-converting copy, and build page layouts.
The Core Idea: Structured Operational Flows
Instead of typing random prompts and hoping for good results, I built **AGENC-IA**.
It is my personal framework consisting of **15 custom-configured AI assistants (Skills)**.
I connect them in a logical chain so that the output of one step serves as the database and reference source for the next.
This ensures that the brand voice, keywords, and marketing goals determined at the beginning are perfectly preserved and used in everything we build later: from the emails and SEO articles to the video scripts and landing pages.
Explore the Modules
Market Intelligence
Deep competitor analysis and avatar mapping
Module 01
Why it works & What we do:
We start by analyzing the market to make sure we aren’t guessing. We conduct keyword research to see what users are looking for, audit competitor websites to see what is working for them, and map out the target audience’s psychological profile—analyzing their deepest fears, desires, frustrations, and goals.
Core Elements Mapped:
- Comprehensive Keyword Audit
- Competitor Analysis & Site Audits
- Avatar Fears, Desires, & Goals mapping
- 100-Topic Content Calendar Ideas
- Avatar Digital Channel Map
- Structured 10-Point Core Questionnaire
– Role: Perform deep competitor and target audience research.
– Input: Ingest seed business details and target audience profile.
– Operations:
1. Audit competitor assets (find traffic, keywords, positioning gaps).
2. Map out avatar psychological triggers:
– Fears (What keeps them awake at night?)
– Desires (What is their dream outcome?)
– Frustrations (What is stopping them?)
– Goals (What are they trying to achieve?)
3. Generate 50 high-intent keywords + 100 content topics for AEO/SEO indexing.
– Output: Standardized markdown format saving all avatar and keyword data to feed Module 2.
Value Offer Architect
Structuring services into high-value bundles
Module 02
Why it works & What we do:
Instead of selling generic services, we package them into an «irresistible offer.» We apply **Alex Hormozi’s Value Equation** (maximizing dream outcome and probability of success, while minimizing time delay and effort) to stack custom bonuses and guarantees that reverse the buyer’s risk. Then, we structure a logical **Value Ladder** (inspired by **Russell Brunson**) to guide the customer from low-ticket entry points to high-ticket backend services.
Core Elements Mapped:
- Alex Hormozi Value Equation Calculations
- Russell Brunson Value Ladder Stacking
- Core Service Restructuring & Bundling
- Risk-Reversal & Guarantees
- Low, Mid, and High Ticket Tier Structuring
- Urgency & Scarcity Copy Triggers
– Methodology: Ingests principles from Hormozi’s $100M Offers and Brunson’s Value Ladder.
– Input: Reads the psychological profile and competitor gaps from Module 1.
– Operations:
1. Calculate value: (Dream Outcome * Perceived Likelihood) / (Time Delay * Effort & Sacrifice).
2. List user obstacles and convert them into valuable bonuses (Bonus Stack).
3. Build the Value Ladder:
– Lead Magnet / Bait.
– Front-end Offer (Low ticket).
– Middle-tier Offer (Core service).
– Back-end Offer (High ticket / Retainer).
4. Design a risk-reversal guarantee directly addressing the client’s biggest fear.
– Output: Structured offer copy briefs to feed Module 3 (Copy Engine).
Semantic Copy Engine
Brand voice definition and copywriting library
Module 03
Why it works & What we do:
Also known as Kopy Krafter. We standardize the brand’s tone of voice and write the core marketing hooks.
We map the target keywords identified in Module 1 and select exactly where to use them in the copy (headers, subheaders, and body) to ensure perfect SEO and conversion alignment.
Core Elements Mapped:
- Tone of Voice Guidelines
- StoryBrand Narrative Framework
- Keyword Mapping & Copy Placement
- 26-Hook Dynamic Copy Library
- Core Founder’s Story Script
- Semantic Coherence Standards
– Input: Uses the offer details (Module 2) and target keywords (Module 1).
– Operations:
1. Map out target keywords and define exactly which terms go in Headlines (H1), subheaders, and CTA copies.
2. Define the Brand Voice (Tone, words to avoid, style guidelines).
3. Draft the client journey using Donald Miller’s StoryBrand blueprint.
4. Generate a hook library (26 distinct hooks based on Eugene Schwartz awareness levels).
– Output: Standardized Copy Brief to feed website builders and content factories.
SEO Content Factory
Scale article generation without quality loss
Module 04
Why it works & What we do:
We create in-depth, high-authority articles using Article Krafter. The articles are structured specifically to answer questions directly (Q&A sections) and include structured schema markup. This makes it incredibly easy for Google and generative AI engines (ChatGPT, Claude, Perplexity) to find, extract, and cite your brand.
Core Elements Mapped:
- Helpful Content SEO Audits
- Structured Data Markup (JSON-LD)
- Direct Q&A Blocks for AI Crawlers
- Strategic Semantic Keyword Insertion
- Internal & External Anchor Layouts
- Readability & Scannable Layout Formatting
– Input: Reads semantic copywriting rules and keywords from Module 3.
– Operations:
1. Write structured long-form articles (1,500+ words).
2. Implement direct question-and-answer headings:
– Use H2/H3 for common queries.
– Provide a clean, direct 2-sentence response immediately below for AEO/GEO snippet extraction.
3. Insert structured JSON-LD schemas.
– Output: Publish-ready markdown file optimized for both Google and AI search engines.
Video Prompting / VSL
Using creative assets as targeting signals
Module 05
Why it works & What we do:
In modern paid traffic (like Meta Ads), **your creative is your targeting**. Instead of manually selecting interest groups, we write video scripts structured specifically to act as a **»human prompt»** for Meta’s algorithm.
By using hooks and language that only speak to our target avatar, uninterested users scroll away immediately, while target buyers stay hooked. Meta’s AI tracks this retention, and uses its **500+ user data points** to automatically optimize delivery and find similar buyers. This performs far better and scales easier than old-school manual interest targeting.
Core Elements Mapped:
- 12-Step Structured VSL Formula
- Algorithmic Hook & Scroll Stopper design
- Self-Qualifying Hook scripts
- Value Stack & Bonus Scripting
- Visual & On-Screen Cue directions
- Pacing and Audience Retention Guidelines
– Input: Ingests the Irresistible Offer stack (Module 2) and Brand Voice hooks (Module 3).
– Operations:
1. Design a VSL script using a 12-step hook-story-offer blueprint.
2. Implement «Creative Targeting» Hooks:
– The first 3 seconds must explicitly state the avatar’s specific problem (e.g., «If you install artificial turf…»).
– This filters out non-buyers and feeds Meta’s algorithm with clean user-retention signals.
3. Script the Value Stack and Bonus delivery.
4. Embed visual cue directions: [Visual: Show screen demo], [Tone: Conversational].
– Output: Production-ready video script draft.
Landing Page Engineer
High-conversion landing page layouts
Module 06
Why it works & What we do:
Also known as Landing Krafter. We take all the work done in the previous steps—the buyer avatar, the irresistible offer, the copywriting hooks, and the video script—and translate them into a clean, mobile-responsive page layout designed to convert visitors into leads.
Core Elements Mapped:
- Conversion-focused Grid layout specs
- Above-the-Fold (Hero Section) structure
- Video & VSL Embed Area styling
- Bonus Grid & Value Stack visuals
- Risk-Reversal & Guarantee card styles
- Mobile-Responsive Layout guidelines
– Input: Reads copywriting assets (Module 3) and offer structures (Module 2).
– Operations:
1. Map the landing page wireframe sections in sequence:
– Header & Logo.
– Hero Section (Headline, Subheader, VSL video embed, CTA button).
– Social Proof Grid (reviews, logos).
– The Value Stack Grid (displaying the core service + bonus cards).
– Guarantee Card (Risk reversal).
– FAQ Accordion (containing AEO question-and-answer snippets).
2. Standardize spacing rules, font hierarchy, and mobile layout breakpoints.
– Output: Frontend layout specification for web developers.
AI Media Factory
Consistent image & video asset generation
Module 07
Why it works & What we do:
To fuel our ads and websites, we generate high-quality images and video ads using **Higgsfield** (for video) and **Magnific AI** (for image upscaling). To keep this process fast and automated, I connect to these APIs via **MCP (Model Context Protocol)** servers directly inside my workspace.
To maintain absolute visual consistency across generations (avoiding deformed products or changing characters), I built a structured directory framework. The AI reads raw reference images and paired **JSON data files** representing: the **Design System** (styles/fonts/colors), **Character Sheets** (describing characters from all viewing angles), **Product Sheets** (describing products from all viewing angles), and **Environment Sheets** (setting the background/atmosphere).
Core Elements Mapped:
- AI Video Generation (Higgsfield via MCP)
- Image Enhancements (Magnific via MCP)
- Multi-angle Product Sheets (JSON + Refs)
- Multi-angle Character Sheets (JSON + Refs)
- Obsidian Design System & Environment guides
- Structured Asset Directory Schemas
– Role: Coordinate and automate image and video generation pipelines.
– Folder Structure:
– `/design-system/`: Theme parameters, colors, visual rules.
– `/characters/character_sheets.json`: Details on characters + reference images (must include Front, Profile, 3/4, and Back angles).
– `/products/product_sheets.json`: Geometry details, labels + reference images (must include Top, Bottom, Side, and Angled closeups).
– `/environments/environments.json`: Detailed descriptions of lighting, spaces, and backgrounds.
– Operations:
1. Ingest target scene prompt from Module 3 Copy Brief.
2. Parse the character, product, and environment sheets from the local JSON files.
3. Format the API call to Higgsfield/Magnific incorporating character/product reference images as control signals.
4. Execute API calls via MCP tools to output the final upscaled image/video.
– Output: Standardized assets saved directly to the site deployment directory.
How I Set Up This System
This system runs smoothly because it is organized as a **Markdown-based database** inside my local environment:
1. Markdown File Datastore
Every step of the framework saves its variables, keywords, and draft results to a structured `.md` text file. I manage these files in a local folder hierarchy. This acts as the database state that allows my AI agents to reference previous steps accurately without context loss.
2. Custom System Instructions (Skills)
Each agent has a set of markdown-formatted system instructions (Skills). These files dictate what inputs the agent must read, what rules it must follow (like avoiding typical marketing clichés), and what schema formatting it must output.
3. Copywriting Keyword Injection
By passing the keyword map from Module 1 directly into the copywriting prompts in Module 3, I ensure that all headlines, landing pages, and body copies are automatically optimized for target keywords naturally.
4. Continuous Refinement
I version-control my prompts and skills files. When a campaign finishes and I analyze the metrics (conversion rates, search rankings), I update the rules in the skill files so the AI team gets sharper and faster with every project.