AETHERIS & THYMOS
How It Works: Adversarial processing
Instead of a linear script, Aetherya simulations operate on a Dynamic Tension Model.
- The Plan: CORTEX analyzes the page and determines the "logical" next step (e.g., Click 'Sign Up').
- The Resistance: THYMOS evaluates the friction of that step against the persona's current emotional state. Is the button hard to find? Is the text dense?
- The Resolution: The action is executed based on the collision of these two forces.
- Low Friction: The agent acts smoothly.
- High Friction: The agent may hesitate, mis-click, or—if the frustration threshold is breached—abandon the journey entirely.
This interplay ensures that when an Aetherya agent fails to convert, it isn't a bug in the code—it is a valid prediction of where your real users will struggle.
Aetheris Cortex: The Reasoning Core
Cortex is the strategic intelligence layer. When you ask a persona to "evaluate this checkout flow," Cortex is responsible for understanding the instruction, breaking it into logical steps, maintaining context across multiple pages, and constructing a coherent narrative of the experience.
Unlike standard language models that simply generate text, Cortex operates as a reasoning engine. It doesn't just respond—it plans, evaluates, and adapts. This is what allows personas to navigate complex multi-step journeys, remember previous interactions, and make decisions that are logically consistent with their established goals and constraints.
Cortex handles:
- Multi-step logic: Breaking down complex tasks into sequential actions
- Context retention: Remembering what happened 10 pages ago and how it influences current decisions
- Goal alignment: Ensuring every action serves the persona's defined objectives
- Strategic planning: Determining the optimal path through an interface based on persona priorities
In practice, this means a persona doesn't just click randomly. It evaluates options, weighs trade-offs, and makes decisions that reflect a coherent internal strategy—just like a real user would.
THYMOS: The Human Factor
THYMOS is what makes Aetherya unique. It's a proprietary machine learning engine that introduces the human factor—the emotional, cognitive, and behavioral variability that defines real-world user behavior.
Standard AI simulations fail because they're too perfect. They don't get frustrated. They don't misread instructions. They don't abandon a form because they're tired or distracted. THYMOS fixes this by layering human-like variability on top of Cortex's logical reasoning.
THYMOS modulates behavior across several dimensions:
- Emotional state: A persona experiencing friction will become impatient, leading to faster, less careful interactions
- Cognitive load: Complex interfaces increase mental fatigue, reducing attention to detail and increasing error rates
- Attention span: Personas can become distracted or lose focus, especially in long, repetitive flows
- Behavioral entropy: Real users don't follow perfect paths—they backtrack, hesitate, and make mistakes. THYMOS replicates this
The result is simulation data that mirrors real user behavior. When a persona abandons a checkout, it's not because of a random dice roll—it's because the cumulative friction exceeded their tolerance threshold, just like a real customer.
How They Work Together
Cortex and THYMOS operate in tandem. Cortex provides the what and why of a decision, while THYMOS provides the how—the messy, human execution of that decision.
For example, when a persona encounters a confusing form field:
- Cortex recognizes the field is required and determines the logical next step is to fill it out
- THYMOS introduces hesitation—the persona pauses, re-reads the label, and may enter incorrect data on the first attempt
- Cortex evaluates the error message and decides to correct the input
- THYMOS increases frustration, making the persona more likely to abandon if another error occurs
This interplay creates behavior that is both strategically coherent (Cortex) and behaviorally realistic (THYMOS). It's the difference between a simulation that tells you "users might struggle here" and one that shows you exactly how and why they struggle.
Cognitive Mode: Deep Reasoning
When you enable Cognitive Mode in an audit, you're unlocking the full depth of the Cortex reasoning engine. This mode applies a 1.8× session multiplier because it requires significantly more computational resources—but the insights are worth it.
In Cognitive Mode, personas don't just react to what they see. They:
- Analyze the intent behind design choices
- Compare experiences across multiple pages to identify inconsistencies
- Evaluate whether messaging aligns with their expectations and mental models
- Provide strategic recommendations based on their persona archetype
This is the mode you use when you need to understand not just what users do, but why they do it. It's the difference between "users dropped off at step 3" and "users dropped off at step 3 because the value proposition became unclear and the cognitive load of the form exceeded their patience threshold."