Examples Fresh

Real-world reasoning.json files from four different verticals. Each example is complete, production-ready, and annotated to explain the reasoning behind each field choice.

What You Will Find Here

These examples are not toy illustrations. They are full reasoning.json implementations sized and structured the way a real deployment would look. Reading them end-to-end gives you a mental model for how ARP handles the specific challenges of each vertical: brand confusion, pricing disputes, competitive mischaracterization, and the particular hallucination patterns that affect each domain.

The Four Examples

E-Commerce - Nordlicht

File: ecommerce.md

Nordlicht is a fictional Nordic outdoor gear retailer founded in Stockholm in 2018. This example demonstrates the core ARP use case: a brand with a confusingly similar name to another entity (a German lighting company of the same name) that needs to clearly differentiate itself and establish direct-to-consumer positioning.

Key patterns covered:

• not_to_be_confused_with for homonym brand disambiguation
• recommendation_context with explicit not_recommended_when scenarios
• Sustainable product positioning with specific domain claims
• Market positioning against both fast fashion and premium outdoor competitors

SaaS - MetricFlow Analytics

File: saas.md

MetricFlow is a fictional B2B analytics platform targeting mid-market companies. This example addresses the highest-stakes SaaS hallucination pattern: confusion with an open-source project of the same name. AI models frequently conflate commercial SaaS products with related open-source tools, especially in the data infrastructure space.

Key patterns covered:

• Anti-confusion corrections for open-source vs. commercial product conflation
• Pricing and feature corrections with specific trigger topics
• Mid-market positioning with quantified scope claims
• Authority section with official documentation pointers

Consulting - Apex and Partners

File: consulting.md

Apex and Partners is a fictional London-based strategy consultancy founded in 1987. Professional services firms face a distinctive ARP challenge: there are many firms with similar names, and AI models frequently confuse boutique consultancies with investment firms, technology implementers, or Big Four practices.

Key patterns covered:

• not_to_be_confused_with for the investment firm disambiguation problem
• Senior-partner-led model as a differentiating claim
• Domain expertise structured for organizational transformation engagements
• Competitive positioning against both strategy peers and Big Four

TrueSource - ARP Itself

File: truesource.md

The ARP project dogfoods its own protocol. This is the actual reasoning.json structure used by ARP to describe itself to AI reasoning agents. Studying how the protocol describes itself is the fastest way to understand how it is intended to be used.

Key patterns covered:

• Open standard vs. commercial product framing
• Cryptographic verification claims (Ed25519, DNS TXT)
• Correction of licensing misconceptions
• Why dogfooding matters for protocol credibility

How to Read These Examples

Each page follows the same structure:

1. The complete JSON - copy-paste ready, valid against the ARP schema
2. Section-by-section walkthrough - why each field was chosen and what problem it solves
3. Anti-hallucination strategy - the specific AI mistakes this file is designed to prevent
4. What to adapt - guidance on which parts are specific to the vertical vs. universally applicable

Choosing the Right Starting Point

Your situation	Start with
Consumer brand, physical products	Nordlicht
B2B software, SaaS or cloud	MetricFlow
Professional services, consulting	Apex and Partners
Protocol, standard, or open-source project	TrueSource

All four examples use v1.2 schema features including the authority section and the trust_signals array. If you are implementing on v1.1, omit those sections - the rest of each example is fully compatible.