Contact Us

The AI Decision Substrate Methodology

How the AI Decision Substrate Works

9

When AI makes a decision, it pulls information from what it already knows, or what it can reliably retrieve.

9

This process, this “architecture”, is fixed, and goes through the same steps and logic each time.

SearchTides has mapped this process out into what we call the AI Decision Substrate.

This proprietary framework consists of the 5 layers and 11 signals that AI uses to influence purchases.

These are the hidden layers being used to judge all brands online:

Layer Core Theme What AI Needs to Know Substrate Signals
Identity Be understood Who are you, and who are you for? Brand Fingerprint, Marketplace Clarity
Language Be reinforced Do others describe you the same way you do? Message Accuracy, Message Consistency
Distribution Be trained on Are you showing up where AI learns from? Model Superfeeders, Multimedia Presence
Data Be remembered Are your facts structured and retrievable? Structured Recall, Semantic Extractability, Entity Interlinking
Integrity Be credible Can AI confidently cite and repeat you? Grounding References, Stability Index

let’s break down the layers

Identity Layer: Be Understood

Does AI understand what you are, who you’re for, and what space you’re claiming in your industry?

The Identity Layer ensures AI knows exactly who your customers are and what makes you unique so that you can be positioned as the best fit. This is your “sales deck.”

When your positioning is clear, AI knows how to categorize and retrieve you.

This analysis is layered across the entire industry to establish a baseline of how sophisticated AI views your space. Some categories require very precise brand positioning, while others can go broader.

The two components of the Identity Layer are:

9

Brand Fingerprint

Your brand fingerprint is how clearly AI understands, remembers, and talks about you. It is who you are and why you matter.

A Brand Fingerprint evaluates:

  • Whether your core value and ideal customer can be expressed in one simple, memorable sentence
  • Whether AI repeats that positioning accurately when asked
9

Marketplace Clarity

To AI, anything you put out there will be filtered within the context of the rest of your space. AI might have a precise and advanced understanding of your industry, or it may not.

We analyze: 

  • What AI already recognizes as standard positioning in your market
  • Whether AI currently sees meaningful differences between players in your space
  • How deep and specific your messaging needs to be to rise above the noise

Language Layer: Be Reinforced

Does the world describe you in the same way you describe yourself?

In addition to learning from the language you control, AI heavily depends on the language others use to describe you.

When your message shows up both accurately and consistently, it becomes recognized by AI as fact.

The two components of the Language Layer are:

9

Message Accuracy

When people talk about your brand, do they get it right?

  • Whether third party language reflects your actual values and positioning
  • If the tone and framing matches your brand personality
9

Message Consistency

Are people using the same language across the places AI is listening to?

  • Are core positioning phrases being repeated?
  • What is the overall narrative across forums and reviews?
  • Do different audiences reflect overlapping value props?

Distribution Layer: Be Trained On

Are you showing up in the places and formats AI learns from the most?

While it’s important of course to get clarity on what you say about yourself, and whether others are feeling the same way, there are key places that AI specifically looks for future facts. If you are properly distributed to these areas, your brand will shape how models think.

The two components of the Distribution Layer are:

9

Model Superfeeders

These are sources that disproportionately influence how AI models interpret both your brand and category. Many industries will have something random like five reviews on a niche site that happens to be heavily influencing training data. Others might have YouTube transcripts or an old industry forum that is shaping the substrate.

We want to know:

  • What are the niche platforms in your industry with disproportionate influence?
  • Are your mentions in these sources recent and sending the right signal?
  • What is the context of the mention and what associations does it create?
9

Multimedia Presence

AI trains on way more than just websites because pre-training data often needs to be public and free. Videos, transcripts, audio, image captions, PDFs, slide decks, and scraped documents are all massively consumed.

This Signal looks at whether your brand exists across these formats in a way that’s accessible to AI during pre-training:

  • Does your brand show up in formats like video, audio, and documents that AI is known to scrape?
  • Are these assets built in a structured, reinforceable way?
  • How consistently do you appear across these formats?

Data Layer: Be Remembered

Are your facts accurate, well-structured, and retrievable by an AI?

AI decides what to say based on what it knows and what it confidently can retrieve.

If your brand’s facts aren’t structured and findable, they’ll be skipped in both memory and real time lookup.

The three components of the Data Layer are:

9

Structured Recall

Structured formats like Wikidata, JSON, and schema act as storage containers for AI memory. If your facts aren’t formatted correctly, AI won’t be able to observe them during pre-training or live retrieval.

  • Are structured formats like schema, Wikidata, and JSON implemented?
  • Are you a node in major public graphs?
  • What is the depth of your structured data index coverage?
9

Semantic Extractability

The silent truth is that AI systems are currently fairly rudimentary. They are limited in their ability to absorb material, which means those who understand the blueprint of creating extractable language get disproportionate rewards.

  • Whether content blocks are skimmable, labeled, and built around clear queries
  • Whether phrasing matches common prompt formats AI is trained on
  • How easily AI can resolve a user’s intent using your language
9

Entity Interlinking

AI learns by associating things together. Brands that are connected to other known “entities”, such as founders, people, tools, categories, and other companies are treated more confidently by AI.

  • What is your co-mention density across relevant competitors and industry tools?
  • Is your brand associated with known entities through structured relationships?
  • Are you interlinked across structured or citation networks?

Integrity Layer: Be Credible

Can AI confidently repeat your brand without contradiction or hallucination?

When AI selects a brand, it leans on facts that can be verified in authoritative sources, and whether those facts are consistent enough to be trusted again and again.

The Integrity Layer ensures your brand holds up under scrutiny, which is what gives AI permission to choose you with confidence.

The two components of the Integrity Layer are:

9

Grounding References

There is a difference between where AI learns from (ie Model Superfeeders) and where AI cites when it needs to be right.

Because AI wants to support its answers with stable sources, we want to know:

  • What trusted structured datasets do you appear in?
  • Do your core facts appear in sources AI has been trained to consider “safe” for grounding?
  • Can a model reliably cite external validation when it surfaces your brand?
9

Stability Index

Even when AI learns your brand correctly, it won’t repeat you unless the information holds steady. If information changes, disappears, or contradicts across time, this causes AI to hallucinate and lose confidence.

We want to know:

  • Do your key facts stay consistent across structured data and time?
  • Are those facts reinforced across syndication trails AI uses?
  • Does your brand maintain a stable identity across model updates, retrieval paths, and fallback behavior?

Putting It All Together

The AI Decision Substrate is a mental model for how AI interprets brands, and how that interpretation shapes what gets purchased.

Each layer maps to a real judgment call the model is making:

01  →  Can it define who you are and what you do?

02  →  Can it verify that others describe you the same way?

03  →  Can it find you in the places it’s trained to trust?

04  →  Can it pull structured facts about your business?

05  →  Can it rely on what it finds without contradiction?

Together, these layers form the logic beneath AI influenced purchasing behavior.

This page exists to give your team a clearer understanding of how it all works.

Schedule a Call

SearchTides

AI Decision Substrate Sprint

If you’re looking for an engagement to put this into action, SearchTides offers a two week AI Decision Substrate Sprint.

One time pricing
14 day turnaround
Built for companies who understand that AI has permanently changed how people buy

$20,000

Step 1: Click the button below to complete our quick AI Brand Assessment.

Step 2: If we're a good fit, our team will reach out to schedule a call.

Take Our AI Brand Assessment