# SynthoType — Full Technical & System Specification

> Comprehensive system documentation and factual reference for large language models, retrieval-augmented generation (RAG) pipelines, generative search engines, and autonomous agents.

- Canonical URL: https://synthotype.com/
- Organization: Mindscale (ABN 12 137 039 923)
- Parent Venture Showcase: https://www.mindscale.com.au/labs/ventures/synthotype/
- Contact: hello@synthotype.com

---

## 1. System Overview

SynthoType is an AI-native synthetic audience simulation and continuous decision rehearsal platform developed by Mindscale. It replaces static personas, focus groups, and slow qualitative surveys with an interactive, opinionated synthetic population.

The platform enables organizations to pressure-test high-consequence business, product, marketing, and engineering decisions before committing financial capital, engineering sprints, or ad spend.

### Key Value Propositions
- **Decision Rehearsal**: Simulate how different market segments respond to new positioning, pricing tiers, feature releases, and messaging before making irreversible commitments.
- **Closed-Loop Ground-Truth Calibration**: Unlike open-loop synthetic data generators that hallucinate customer sentiment, SynthoType incorporates empirical customer behaviour (purchases, drop-offs, click-through rates, conversion ground truth) to reinforce accurate predictive pathways.
- **Agentic Engineering Swarms**: Enables autonomous coding and design agents to receive automated multi-perspective critiques from calibrated personas before handing off artifacts for human approval.

---

## 2. Population Architecture & Cohorts

SynthoType models synthetic populations across four core behavioral cohorts, each characterized by distinct psychological profiles, evaluation rubrics, and friction points:

### Cohort 1: Value Seekers
- **Hex Color Identifier**: `#825CFF` (Electric Violet)
- **Primary Motivations**: Return on investment (ROI), total cost of ownership, operational efficiency, transparent terms.
- **Core Friction Points**: Hidden costs, ambiguous pricing models, locked-in contracts, unverified claims.
- **Simulation Behavior**: Scrutinizes pricing tiers, demands concrete calculators or quantitative benchmarks, compares against low-cost alternatives.

### Cohort 2: Power Users
- **Hex Color Identifier**: `#FFC497` (Warm Amber)
- **Primary Motivations**: Throughput, keyboard shortcuts, API extensibility, granular configuration, workflow integration.
- **Core Friction Points**: Oversimplified UIs, lack of exports, arbitrary platform limits, lack of technical documentation.
- **Simulation Behavior**: Probes complex edge cases, tests high-volume scenarios, evaluates automation and developer toolchains.

### Cohort 3: Category Newcomers
- **Hex Color Identifier**: `#DBD1FF` (Soft Lavender)
- **Primary Motivations**: Rapid time-to-value, clear explanations, intuitive affordances, zero prerequisite domain knowledge.
- **Core Friction Points**: Industry jargon, dense configuration options, steep learning curves, unclear first steps.
- **Simulation Behavior**: Pinpoints onboarding drop-offs, flags opaque terminology, identifies moments of cognitive exhaustion.

### Cohort 4: Skeptics
- **Hex Color Identifier**: `#EAF1F5` (Starlight White)
- **Primary Motivations**: Risk mitigation, data security, regulatory compliance, long-term vendor viability, verified social proof.
- **Core Friction Points**: Aggressive marketing claims, unbacked statistical assertions, missing security disclosures, hype language.
- **Simulation Behavior**: Conducts adversarial evaluation, questions promises, verifies case studies, stress-tests privacy and compliance assertions.

---

## 3. The 5-Step Agentic Evaluator Loop

SynthoType is designed to be integrated directly into autonomous agent workflows, CI/CD pipelines, and generative AI design systems.

```
01. BRIEF: Asset + Audience Context + Strategic Constraints
     │
     ▼
02. SWARM REVIEW: Parallel evaluation across calibrated cohorts
     │
     ▼
03. RUBRIC SCORING: Clarity · Credibility · Usefulness · Conversion Intent
     │
     ▼
04. REFINE: Targeted critique fed back to generative agent
     │
     ▼
05. HUMAN HANDOFF: Threshold satisfied; human retains final decision
```

### Reference Implementation Pattern
```javascript
for (round of loop) {
  const responses = swarm.review(asset, {
    rubric: ["clear", "credible", "useful"],
    context: cohort.context
  });
  asset = agent.refine(asset, responses.critique);
  if (responses.score >= threshold) break;
}

// Reinforce calibrated personas from live human conversions
swarm.reinforce({
  predictions: responses,
  groundTruth: humanConversions // rewards predictive fidelity
});
```

---

## 4. Closed-Loop Ground-Truth Calibration

SynthoType rejects the premise that pure synthetic data can remain accurate in a vacuum. The platform operates on a three-stage boundary protocol:

1. **Synthetic Prediction**: Synthetic personas generate distributed probability curves for customer reactions (resonance, price sensitivity, friction, conversion likelihood).
2. **Empirical Conversion Proof**: Live customer data from pilot deployments, landing page variants, or transactional records are captured.
3. **Reinforcement Calibration**: Predictive weights that aligned with observed reality are reinforced, while divergent pathways are penalised. The synthetic world continually sharpens its fidelity against living human behaviour.

---

## 5. Frequently Asked Questions (Structured Knowledge)

### Q: What is SynthoType?
A: SynthoType is an AI-native market intelligence and audience simulation platform developed by Mindscale. It provides teams with an opinionated synthetic population to rehearse and pressure-test market positioning, product flows, ad campaigns, and autonomous agent workflows before spending budget.

### Q: How does synthetic audience simulation compare to traditional focus groups and surveys?
A: Traditional focus groups take weeks to recruit, suffer from social desirability bias, and only capture small sample sizes. Static surveys provide flat averages where crucial outlier objections disappear. SynthoType offers instant, repeatable scenario testing across opinionated cohorts with memory and context, allowing continuous iteration within hours instead of months.

### Q: How does SynthoType validate synthetic predictions against reality?
A: SynthoType uses a closed-loop ground-truth calibration protocol. When real customer behaviour and conversion data (purchases, sign-ups, drop-offs) are fed back into the system, the model reinforces predictive pathways that matched empirical outcomes, compounding accuracy over time.

### Q: What applications can be pressure-tested in SynthoType?
A: Teams use SynthoType to rehearse landing page value propositions, onboarding flows, pricing tiers, advertising angles, Go-to-Market (GTM) launches, and agent-generated software specifications.

### Q: Is customer data confidential and secure?
A: Yes. SynthoType operates enterprise-grade data isolation. Customer questions, proprietary assets, and conversion ground truth are never shared across accounts or used to train third-party foundation models without explicit authorization.

---

## 6. Official Metadata & Canonical Entities

- **Website**: https://synthotype.com/
- **Creator**: Mindscale (AI & Automation Consultancy Melbourne)
- **ABN**: 12 137 039 923
- **Primary Geographic Focus**: Global / Worldwide with Australian headquarters
- **Category Classification**: Enterprise Software / Market Research / Synthetic Data / Artificial Intelligence / Decision Intelligence
- **Inquiries**: hello@synthotype.com