Decisions Shouldn't Vanish
When Tools Change
IRP is a protocol for capturing, validating, and querying decisions across your organization. This book explains the architecture, the design patterns, and how to apply them to your own decisions. Built from first principles. No vendor lock-in.
Start readingWhat you'll learn
Ledger as source of truth
Why JSONL instead of a database. Why append-only is the only safe design. How immutable logs make decisions portable.
Derived state & conflict detection
Computing current.json from a ledger. Heuristic-based conflict detection without embeddings. Why non-blocking validation prevents friction.
Multi-source capture
Interactive capture, stdin automation, sensor architecture. How Figma, Slack, and CLI all write to the same ledger without sync problems.
The bridge pattern
Three-layer architecture for tool integration. Why sandboxed environments need a bridge. How to add new sensors without touching the core.
REST APIs & extensibility
/inherit, /why, /check endpoints. Context injection for external AI models. Cross-tool decision workflows. Sovereignty through local storage.
Real-world scaling
Team sizing (10 vs 100 people). Multi-repository setups. Ledger maintenance, archiving, and backups. When IRP matters and when it doesn't.
A Plain-Language Guide to EU AI Act Compliance
The EU AI Act doesn't regulate AI. It regulates whose hands are on the steering wheel — and whether you can prove it. Nine chapters cover the law in plain language: why it exists, how to identify your role (Provider, Deployer, Importer), and what each key article actually requires you to document. Article 12 mandates an audit trail that saves prompts to a database won't satisfy. Article 14 demands evidence that a human was in control. Article 72 turns compliance into a permanent ongoing obligation. Every obligation maps back to the same thing: durable, portable records of decisions made and why — which is exactly what a plain ledger provides.
Who this is for
Teams building decision management systems. Every chapter ends with "Apply This" — transferable patterns with concrete adaptation advice. Use the patterns, not the code.
Architects evaluating durability patterns. Learn why ledger-as-source-of-truth matters. Understand the tradeoffs between speed, explainability, and cost.
Anyone tired of decisions disappearing. When teams change tools, decisions die. IRP makes them survive. Apply the patterns to your own decisions.
Table of contents
Core Framework
A decision is an artifact. Make it durable.
Why decisions vanish, the ledger-as-source-of-truth pattern, and immutable audit logs
Derived state, the check algorithm, lightweight heuristics, and non-blocking validation
Interactive and stdin capture modes, sensor architecture, and context enrichment
Integration & Extension
Decisions live everywhere. So should their ledger.
The check command, keyword-based conflict detection, and team resolution
Three-layer architecture, bridge pattern, message-based relay, and conflict checking integration
REST APIs, MCP protocol for agents, context injection for AI models, sovereign stack integrations (Obsidian + MemPalace), and adding new sensors
Application & Practice
The patterns are universal. Your decisions are not.
A Plain-Language Guide to EU AI Act Compliance
The EU AI Act doesn't regulate AI. It regulates whose hands are on the steering wheel.
The incidents that moved legislators, what the governance gap actually was, and why the AI Act is different from GDPR
Provider, Deployer, Importer — the role distinction that changes everything, and why "we just use AI" is no longer safe
Four tiers from Unacceptable to Minimal, Annex III high-risk categories, and how to self-assess before the regulator does
What logging means under the AI Act, why saving prompts to a database does not qualify, and what a compliant audit trail looks like
Why every high-risk AI system needs an Instructions for Use document, who writes it, and what happens when it is missing
What human oversight means legally vs operationally, how to prove a human was in control, and the documentation that survives an audit
What a Fundamental Rights Impact Assessment requires, who must conduct one, and why public-sector deployers will be audited on this first
Post-market monitoring as an ongoing obligation, what changes trigger re-assessment, and how to build a compliance system rather than a compliance project
Building compliance as a capability, the compliance stack by company size, and how to use the IRP Compliance Assessment as your starting point
Design philosophy
IRP is built on five core principles: immutability, portability, auditability, explainability, and team autonomy.