PRIVATE BETA · GOVERNED AI

Govern every agent decision.

GRAFOMEM Cloud is the inference-governance layer for agentic AI — every decision logged and signed, every deletion cryptographically proved, policies evaluated before execution, and reports that support compliance. Built on the open GRAFOMEM runtime and the GMP standard.

Verify the memory. Govern the inference. Prove the erasure.

Why Cloud

The open runtime proves your memory. The Cloud governs your whole system.

Self-host the open GRAFOMEM runtime for free — govern memory, sign checkpoints, prove erasure. GRAFOMEM Cloud adds the operational governance layer enterprises need across every inference, at multi-tenant scale.

◇ Core · Open · Self-hosted
"Does my agent's memory behave correctly — and can I prove a thread was erased?"
The Fair-Source runtime + the open GMP standard. Signed checkpoints, erasure receipts, conformance you can run yourself.
◆ Cloud · Managed
"Can I prove my whole AI system is governed, transparent, and compliant?"
Every inference logged and signed, policies evaluated before execution, replayable decisions, and reports that support your regulatory obligations — operated for you, multi-tenant.
The governance layers

Six layers of provable governance over agent execution.

GRAFOMEM Cloud governs your existing agent stack — it doesn't replace your framework. Each layer is cryptographically anchored.

DECISION TRAIL

Every decision, signed

An immutable, Ed25519-signed audit log of every inference — model, query, facts, output, tokens, latency. Nothing an agent decides goes unrecorded.

/v1/decisions/
ERASURE PROOF

Provable deletion

GDPR Article 17 signed erasure certificates. When you delete a subject's data, you get a cryptographic artifact proving the erasure transition occurred — an auditor's evidence, not a promise.

/v1/erasure/
GOVERNANCE GATEWAY

Policy before execution

A policy enforcement point that checks every operation against your rules before it runs — consent, tenant isolation, and access enforced at the gate, not after the fact.

policy · PEP / PDP
REGULATORY REPORTS

Compliance-supporting reports

Generate EU AI Act, GDPR, and DORA reports from the signed record — turning your decision trail and erasure proofs into the evidence a regulator asks for.

/v1/reports/
GOVERNED EXECUTION

Attested workflows

Run agent workflows with a hash-chained attestation on every step, and replay any decision deterministically — governed execution over the framework you already use.

orchestration · replay
VERIFIED MEMORY

Conformant by construction

The managed reference server runs the full GMP memory model on Postgres + pgvector — and passes the same conformance the open standard verifies. No infra, no ops, no eval guesswork.

GMP · M8 = 1.000
Reputation

Verify the agent, not just the answer.

Conformance proves a memory backend does what it claims. Capability-Grounded Reputation (CGR) proves the same for the agents on top of it: each earns a reputation, per domain, from decisions that later resolve right or wrong. It can't be bought, faked, or rushed — it compounds only with real outcomes, and it's Sybil-resistant by construction. Live in the console, early.

RESOLVED OUTCOMES

Grounded in resolved outcomes

A score moves only when a decision is checked against what actually happened — never on assertion, confidence, or volume.

TWO CHANNELS

Verifiable + gated review

Resolved outcomes count directly; peer review counts only as much as a reviewer's calibration, itself earned on verifiable outcomes. Unknown reviewers fail closed.

CONFIDENCE

Confidence travels with the score

Every score carries its evidence mass, so a lucky first call never reads like an established track record.

Built on the open GRAFOMEM standard

GRAFOMEM Cloud runs the same primitives you can self-host for free — the Fair-Source runtime, the `.gfm` format, and the open GMP protocol + conformance suite. Nothing here is locked away: the cloud sells operation, attestation, and audit-readiness at scale — not capability.

Point your registry at the cloud when you want it hosted, governed, and attested. It's a config change, not a migration.

Who it's for

For teams that have to prove their AI is governed.

Financial services
DORA operational-resilience evidence — every AI decision logged, signed, and replayable.
Healthcare
Provable deletion and tenant isolation for patient data across agentic systems.
Public sector & regulated AI
EU AI Act transparency and audit trails, generated from the signed record on demand.
Pricing

Start hosted. Scale into governance.

The runtime is always free to self-host. The Cloud is priced for teams that want it operated, attested, and audit-ready.

Starter
Free
10,000 governed decisions / mo included
The reference server, run for you — for building and early production.
  • Managed GMP memory (Postgres + pgvector)
  • Decision trail + erasure proofs
  • Signed conformance report
Get started
Pro
$20/mo
100,000 governed decisions included, then $0.001 / decision
Full governance for production agentic systems.
  • Everything in Starter
  • Governance gateway (policy before execution)
  • Regulatory reports (EU AI Act · GDPR · DORA)
  • Governed execution + decision replay
  • SSO (Google · Microsoft · GitHub)
Get started
Enterprise
Custom
Regulated deployments, dedicated tenancy, and compliance support.
  • Everything in Pro
  • Dedicated / private-cloud deployment
  • Custom retention & data residency
  • Compliance & onboarding support
Talk to us

Get started in minutes, or log in if you already have an account.

Govern your agents' decisions — provably.

Start free and scale into governance when you're ready — the runtime is always free to self-host; the Cloud is operated, attested, and audit-ready.

Erasure certificates prove a signed state-transition occurred — an auditable cryptographic artifact, not a claim of information-theoretic unrecoverability.