Big News: Kosli’s achieves Series A milestone with Deutsche Bank as an investor - Read the announcement
New: Kosli Answers is here! AI-powered insights for compliance and security. Learn more →

GOVERNANCE INFRASTRUCTURE

What is Governance Infrastructure?

Learn how the infrastructure principles behind Governance Engineering help you record events as verifiable facts, automate controls, and prove compliance continuously.

The Infrastructure Stack

Architecture for Governance.

An architecture for Governance Engineering built on proven technology principles: layered responsibilities, an immutable record, and computed projections. It scales to large, complex environments.

01

Evidence

Record trusted facts from across the SDLC, in a form that can be queried and connected.

An append-only record stores every fact. No actor can fake, modify or delete history.

02

Control

Express and apply policy as controls, mapped to the evidence that satisfies them, operated automatically.

Controls read evidence from the record. Every decision is written back to the record as a fact.

03

Audit

Demonstrate assurance to auditors, regulators, customers, and the board, continuously and as of any point in time.

A digital record of audit data makes audits a query rather than a project. Evidence is available on demand.

04

Insight

Provide insight into how the SDLC and its controls are actually performing.

Use data from across the SDLC to provide status, trends, and metrics, computed from the live record.

The sections below describe each layer in turn. Read on to learn the responsibilities of each layer, and what it looks like in Kosli.

LAYER 01 · EVIDENCE

Record every event as a verifiable fact.

The foundation of governance is a trustworthy system of record. That's why the base of the stack is an append-only, immutable ledger of facts.

CI / pipelines attestations tools / scanners attestations environments snapshots 01 · EVIDENCE append-only immutable record 02 · CONTROL policy as code writes decisions as facts enforcement gates · admission 03 · AUDIT queries · reports · alerts 04 · INSIGHT status · trends · metrics projection projection
Event sources append facts into an append-only immutable ledger.

Attestations and snapshots

ATTESTATIONS: claims that an action happened (tests passed, a change was approved) and SNAPSHOTS: observations of the state of an environment.

Content Addressable Identities

Every artifact is identified by its cryptographic identity such as git commit or SHA256 checksum, not its name or tag. Evidence is connected via identities or correlation ids so that you can build a chain of custody from commit to production.

Append-only, independent record

The system of record has no update and no delete. A correction is a new fact, and the old one remains. Kosli holds the record outside your infrastructure, so altering history isn’t possible.

Flexible evidence types with schemas

Define the fact types your governance needs. Evidence that does not meet its schema is rejected at write time, and evaluated on ingest. Raw material (test output, SBOMs, scan reports) can be attached and fingerprinted in the Evidence Vault.

LAYER 02 · CONTROL

Controls run as code.

The Control layer automates policy as code. Decisions are based on recorded evidence and context inputs.

Event Context Evaluate Evidence Rules Decision Result Triggers Produces Get facts Uses /
An automated control: an event triggers an evaluation of rules against trusted facts, producing a recorded decision.

Controls as versioned entities

Every control has a unique, immutable identity and a version history. Names and descriptions can change; the history does not. Every decision records the control and policy version that made it.

Policy as code

Compliance requirements are versioned, testable policies, evaluated automatically. The results are idempotent: the same input evidence produces the same decision, every time.

Decisions recorded as facts

Compliant or non-compliant, with the reason, the policy version, and the evidence it read. The Control layer’s own actions are written back to the Evidence layer.

Distributed Enforcement Points

Decisions get enforced in CI pipelines, deployment tools, and admission controllers.

Exceptions, overrides, and break-glass

Emergency fixes and approved deviations are part of the system. Each is recorded: who, what, when, and the reason.

LAYER 03 · AUDIT

Audit compliance from a continuous record.

Assurance is monitored from the record. Reconciliation is continuous, and anyone can verify it against the facts at any time.

auditors · regulators · the board reporting engine queries · export systems · departments runtime views changes · policy · coverage as of · 2026-03-03 · 14:02 continuous audit change evaluation alerts / computed from the record 01 · EVIDENCE time →
Reports, runtime views, and continuous compliance evaluation, computed from the record.

Flexible Reporting Engine

Support for queries across controls, organizational entities and systems. Complete evidence bundles for any change: attestations, evidence files, spreadsheets and pdf reports.

Runtime Views

Point-in-time views of how runtime environments evolve, reconciled against change sources, policy evaluations, and control coverage.

Continuous Audit

Compliance evaluation with policies to find non-compliance in real time instead of at audit. Every change is evaluated against policy and evidence requirements.

LAYER 04 · INSIGHT

Data insights into technology governance.

Organizational insight based on facts: the evidence record contains rich information on every deployment, decision, and change. Use it to gain insight into status, trends, and metrics.

engineering · security · leadership deployment frequency lead time trend ↓ control coverage per control · per environment computed from facts 01 · EVIDENCE append-only record time →
Delivery and control measurements are computed from the same record.

Delivery metrics from facts

Deployments are calculated from the evidence, not reported by people. The metrics agree with the audit reports because they come from the same facts.

Control coverage

Track which controls are enforced, compliance and non-compliance rates, and exception processes across the estate.

Getting started

Start with Evidence. Each layer builds on the last.

Adoption follows the dependency order. Controls need evidence, audit needs decisions, and insight needs all three. Each step delivers value on its own.

  1. 01

    Record

    Connect pipelines and environments through integrations, the CLI, or the API. Teams keep their existing tools.

  2. 02

    Control

    Turn existing checks into policies. Report non-compliance first; block when the control is stable.

  3. 03

    Prove

    Produce reports, audit packages, point-in-time reconstruction, and continuous compliance status from the record.

  4. 04

    Improve

    Measure delivery and controls from the same facts, and improve controls with the same methods as software.

Build vs buy

The advantage of an independent evidence store.

The independence requirement decides the build-versus-buy question for the bottom layers.

Security Guarantees

A governance record on your own infrastructure can be altered. Keeping your own governance record is the same as auditing yourself. Passing an audit requires proving the security claims of your system.

Build and maintenance costs

Governance Infrastructure is a complex product, needing extensive development, ongoing maintenance, and guaranteed availability.

Proven technology exists today

Kosli provides Governance Infrastructure, and is running in production at major financial organizations. Kosli maintains the integrations as toolchains evolve, and operates the platform as an enterprise service: with SLA's on availability and strong security guarantees.

Engineer the controls, not the infrastructure

The infrastructure is common across organizations. The controls are not: they encode your obligations, your risk appetite, and your audit history.

Where to apply it

Four entry points. The same four layers.

Each use case is a vertical slice through Evidence, Control, Audit, and Insight, applied to one area of the SDLC.

Build

Software Supply Chain Security

Every artifact accounted for, with a verified chain of custody back to source.

Release

Change Management Automation

Gates that read evidence, not tickets, and a change record that writes itself.

Run

Workload Compliance Assurance

Know what's running, and prove it was approved.

Cloud

Infrastructure Change and Drift Detection

Every change identified by how it was made: through the pipeline or around it.

SOC 2

Type II audited

Tier 1

financial institutions in production

FINOS

SDLC working group co-lead

2019

building since

Get started

See the four layers
on your own pipeline.

We'll show you evidence, controls, and proof on your delivery process, or on ours.

Watch it work, end to end

See the platform enterprise teams use to automate governance across build, release, and runtime.

Kosli in action