Documentation

Foundations · Guide 01

How Rulebook works

The public guide to compiling policies and regulatory obligations, compliance-gating AI decisions, enforcing outcomes and proving what happened.

7 min read

Pre-release documentation. This guide describes the current product design and evaluation build. Packaging, availability and commercial terms may change before release.

01

Compiled rules, not retrieval alone

Enterprise search can find the right-looking passage. A consequential workflow still has to resolve effective dates, conditions, precedence, exceptions and who has authority. Rulebook adds the compliance decision and enforcement layer.

Source documents are compiled into typed proposals. An authorized person reviews the proposal and its evidence. Confirmed rules are then sealed into an immutable, versioned Rulepack that can be queried without asking a model to reinterpret the source every time.

  • Documents remain the source of authority
  • Human confirmation separates proposals from rules in force
  • Published versions can be pinned, compared and replayed
  • A model may extract, plan or explain, but the platform decides
02

The four-part control loop

Every production interaction follows the same operating model.

  • Compile policies, contracts, spreadsheets, procedures and repositories into evidence-linked proposals
  • Reason at the lowest tier that can answer within the caller's ceiling
  • Enforce allow, deny, approval and obligation decisions before protected actions execute
  • Prove the decision with a signed receipt and record every access in the ledger
03

The core safety properties

  • A yes or no conclusion cannot be constructed without confirmed evidence
  • Nothing extracted or installed from a template promotes itself into force
  • A missing fact or ambiguity produces a named needs-review outcome, not a guess
  • Every model-assisted judgment is labelled and recorded
  • T0 remains available without a model in the decision path

Executive briefing

Apply the architecture to one consequential workflow.

Bring one consequential agent workflow. We will map the governing policies and regulations, runtime gate, human authority and auditable evidence path with your team.

Request a briefing