The Space Between Decisions

Ma.ai maps your organisational knowledge — people, companies, contracts, decisions, risks — into a living graph. Every high-stakes decision is examined through four stages of structured reasoning.

Powered by The Crusible. Built for organisations that cannot afford unexamined assumptions.

Already have access? Sign in

01
Input Question
03
Declare Context
05
The Crusible
06
Revelation
Scroll · or read on
Layer 1

Ontology Knowledge Graph

Maps people, companies, contracts, decisions, risks, and events into a typed relationship graph. Grows organically from every decision run.

Layer 2

Failure Library

Logs post-decision outcomes: outcome failure, process failure, assumption failure. Append-only. Each entry links to the original decision and the assumptions the Crusible surfaced.

Layer 3

Context Declaration

Assembles declared context from the graph and failure library. You confirm, modify, or remove each item before the Crusible call proceeds. Mandatory. Auditable.

Layer 4

The Crusible API

Sovereign four-stage reasoning pipeline. Maps, eliminates, pressure-tests, and reveals. Returns a Revelation. Its internal process is unchanged by Ma.ai.

Two outputs. One run.

The boundary between what is proven and what is inferred is explicit, auditable, and the product.

Auditors & Regulators

Correlation Documentation Record

Co-occurrence analysis. Zero causal claims. Structured JSON with hardcoded limitation disclosure. For compliance, due diligence, and legal defence.

Boards & Executives

Operational Narrative

Plain-language summary inferred from correlation. Explicitly labelled as inference. For board presentations, fiduciary communication, and executive briefings.

The difference

The old world left no record.
Ma.ai closes that gap.

Old world

Unexamined decisions

💬
“Should we approve this vendor?”
📧
Email thread. Meeting. Gut feel.
⚠️
Vendor approved. No record exists.Undocumented
✕ No audit trail
✕ Assumptions unexamined
✕ Indefensible
MA.AI —

Examined. Documented. Defensible.

Input question with declared contextStep 01
THE CRUSIBLE
Map → Eliminate → Pressure-test → Reveal
Correlation record generatedSigned
⚠ This is a correlation record, not a causal trace.
✓ Full audit trail
✓ Assumptions surfaced
✓ Board-ready output
Decisions by type

Every high-stakes decision.
One structured process.

⚠ CISO priority

Third-party vendor onboarding. The most common source of enterprise breaches.

Before any vendor is approved, Ma.ai surfaces every assumption your team is making — contractual exposure, integration risk, IR maturity, tool sprawl. The Crusible pressure-tests the decision before it's made.

  • Contractual liability exposure examined
  • IR maturity gaps surfaced
  • Integration failure modes mapped
  • Correlation record exportable to auditors
Should we approve onboarding a new EDR vendor given our current IR maturity, tool sprawl, and contractual liability to our three largest enterprise clients?
Context declared — 6 entities, 2 prior decisions
Map phase complete — 14 assumptions surfaced
Pressure-testing contractual exposure...
Revelation pending
Partial revelation — streaming

Critical assumption exposed: Your IR runbook assumes 4-hour containment. Vendor SLA is 8-hour response. This gap is undocumented and creates material contractual exposure under three client agreements.

Unexamined risk: Tool sprawl across 7 existing EDR-adjacent products creates integration failure probability your team has not quantified...

Sovereign reasoning engine

Built on The Crusible.
Reasoning you can trust.

Ma.ai does not modify The Crusible's internal reasoning. Four structured stages run inside a sovereign external engine. Ma.ai controls what goes in — not what happens inside.

  • Map — entities, relationships, prior decisions
  • Eliminate — obvious answers removed first
  • Pressure-test — assumptions stress-tested
  • Reveal — inferences explicitly labelled
Sovereign · External · Unchanged
The Crusible — Live runPressure-test phase
Should we migrate our customer data workload to AWS eu-west-2 given current GDPR obligations and our three active regulatory audits?
Four stages
MapEliminatePressure-testReveal
Surfaced so far
Assumption exposed: Team assumed GDPR Art.44 transfer applies. Schrems II adequacy decision creates additional exposure not currently documented.
Critical gap: None of three active audits cover cloud-region migration decisions. This creates a new undocumented risk surface.
Prior decision conflict: 2023 architecture review assumed on-prem retention for regulated data. This migration contradicts that decision.
録 Correlation record building...Full audit trail
Enterprise-grade

Audit trail & data governance.

Every decision is declared, documented, and defensible. Your data never leaves your control.

You own the record.

Unlike generic AI tools, Ma.ai produces a signed, append-only Correlation Documentation Record for every decision. Exportable to your auditors, your board, and your legal team. The limitation disclosure is hardcoded — never generated by AI.

  • Append-only failure library
  • Immutable context packages
  • Hardcoded limitation disclosure
  • Org-scoped RLS on all data
🔒

Append-only audit log

Every decision, assumption, and Revelation is immutable. No UPDATE or DELETE on the failure library.

🛡

Org-scoped data isolation

Row-level security on every table. Your graph nodes are never accessible by another organisation.

Declared context gate

Human-in-the-loop confirmation before every API call. Context is signed and immutable after confirmation.

Hardcoded limitation disclosure

Every output carries the disclosure as an app-layer constant. Never AI-generated. Never removable.

📋

Board-ready narrative

Operational Narrative generated alongside the Correlation Record. Compliance-ready on day one.

🔑

AES-256 encryption

Prompt encryption at rest. Supabase service role key never exposed client-side.

Your organisational knowledge
📄
Decisions
🤝
Vendors
📋
Contracts
Risks
👤
People
Living knowledge graph
CrowdStrike EDR onboarding 2024
VENDOR
IR runbook v4 — 4hr containment
DOCUMENT
Client SLA breach — Acme Corp 2023
FAILURE
Architecture decision — AWS migration
DECISION
GDPR Art.44 — Transfer mechanism
RISK
Layer 1 — Ontology

Your organisation's
knowledge, mapped.

Every Crusible run seeds the graph with new entities and relationships. Every failure gets logged. The graph thickens with every decision — making the next Revelation sharper than the last.

  • People, companies, contracts, decisions, risks
  • Relationships typed and versioned
  • Failure library feeds back automatically
  • Ontology mode grounds Crusible in your history
Growing with every decision
Request access

Ready to examine
your decisions properly?

Ma.ai is in closed access. Approved organisations get immediate access to Naked mode — direct Crusible runs with full audit trail. Ontology mode unlocks on Phase 2.

  • No unexamined vendor approvals
  • Every decision defensible to auditors and boards
  • Correlation record from day one

Powered by The Crusible · thecrusible.systems

Active decisions — live
Vendor onboarding — EDR platform
Pressure-testing contractual exposure...
14assumptions surfaced
Architecture decision — cloud migration
Map phase complete. Eliminating...
6entities declared
Regulatory decision — GDPR transfer
Queued · Ontology mode
Board decision — acquisition target
Queued · Naked mode
Ma.ai — The Space Between Decisions