Asplenz Knowledge
Decision Governance for AI-Native Organizations
Structure your decisions.
Let your agents check them.
Govern exceptions.
Trace everything.
Make it operational in CI with Knowledge Verifier.
The problem Asplenz Knowledge solves
AI systems now generate code, modify infrastructure, and refactor architecture.
They move fast.
But they don't know:
- Your architectural commitments
- Your non-negotiable security invariants
- Which exceptions were approved
- What must never change
Without a declared decision layer, AI systems operate on assumptions.
Asplenz Knowledge makes your normative state explicit, structured, and consumable by systems.
The four building blocks
Knowledge organizes your decision surface into four types of entries.
Decisions
Documented architectural choices and strategic commitments.
Immutable once recorded. Preserved with full reasoning and context.
We use PostgreSQL for services requiring ACID guarantees.
Invariants
Absolute constraints that must not be violated.
Security boundaries, compliance requirements, critical policies.
All public API endpoints require authentication.
Rules
Active implementation directives.
Versioned, configurable, mandatory or advisory.
All new services must use structured logging.
Overrides
Governed exceptions to rules or invariants.
Explicitly approved. Scoped. Time-bound when needed. Fully traceable.
Service X temporarily exempt from logging rule until migration is complete.
How systems use Knowledge
Knowledge is designed to be consumed by both humans and machines.
Retrieve the applicable decision state
Systems can request the full normative context for a given scope:
- Active invariants
- Mandatory and advisory rules
- Relevant architectural decisions
- Active overrides
- Explicit relationships
This allows CI pipelines, dashboards, and internal tools to operate with accurate decision awareness.
Evaluate a proposed action
An agent can ask if an action is compliant with the declared rules.
Knowledge returns one of three outcomes:
- Allowed
- Denied
- Requires human approval
Knowledge does not enforce execution. It exposes the evaluation result. Your ecosystem decides how to respond.
Human approval for critical actions
Some constraints require explicit human validation before they can be bypassed.
When an action triggers such a constraint:
- 1The system signals that approval is required.
- 2The designated approver is notified.
- 3The human approves or rejects.
- 4If approved, a scoped exception is recorded.
- 5The action may proceed within declared boundaries.
The agent decides whether to proceed. The human decides whether to allow it. Knowledge records the entire process.
Trace every usage
Every interaction with a declared decision can be recorded:
- Which entry was referenced
- In what context (PR, deployment, documentation)
- Whether it was followed, challenged, or overridden
- By whom
This creates a living map of decision usage across your organization. You gain visibility into compliance coverage, frequently challenged constraints, and patterns of architectural drift.
Real-time decision awareness
When your normative state changes, connected systems can be notified. Agents and CI systems can react immediately. Your teams do not operate on outdated rules.
From declaration to operational governance
Declaring rules is step one. To verify compliance automatically in CI, use:
Knowledge Verifier (Premium Add-on)
Knowledge Verifier analyzes Pull Requests against your declared normative state. It resolves applicable scope, detects violations, validates override coverage, and checks required decision citations.
This transforms Knowledge from a registry into operational compliance infrastructure.
Learn about Knowledge VerifierWhat Knowledge is not
Knowledge is not:
- A wiki
- A code scanner
- A workflow engine
- A policy runtime that blocks execution
- A replacement for engineering judgment
It is a structured decision layer for AI-driven systems.
What Knowledge replaces
| Before | After |
|---|---|
| Architecture decisions in documents | Structured, versioned, queryable entries |
| Security rules in someone's head | Explicit, checkable invariants |
| Exceptions in Slack threads | Governed, traceable overrides |
| "What did we decide about X?" | Structured queryable answer |
| Reviews without decision context | PRs evaluated against declared rules |
Start declaring your decision surface
Govern decisions at AI speed.
Asplenz Knowledge - operational decision governance for AI-native engineering.