Decision Guide · CertifiedData
Which CertifiedData component do you need?
Nine decision branches — from synthetic data certification to AI agent payments to compliance logging. Find the right starting point for your workflow.
How to use this page
- 1.Find the signal that matches your situation below.
- 2.Read the answer — each branch tells you exactly which component to start with.
- 3.Follow the links to the detailed surface for that component.
Nine decision branches
Match your situation to a branch. Each branch is independent — you don't need to read them in order.
"I'm generating synthetic datasets"
Do you need to prove a dataset was synthetically generated?
Certify each dataset at generation time. The certificate records the algorithm, SHA-256 hash, and timestamp — anyone can verify the dataset is synthetic without calling you.
"My AI agent makes payments"
Does your agent execute transactions — buying data, services, or compute?
Agent Commerce handles the full payment lifecycle: policy-gated authorization, execution, and a signed receipt that proves spend was approved before execution. No manual approval loop.
"I'm selling datasets to other agents"
Do buyers need cryptographic proof of what they purchased?
Combine certification + Agent Commerce. Certify each dataset, then sell through a policy-gated payment flow. Buyers receive the file, the certificate, and a signed payment receipt — all independently verifiable.
"I need to log AI decisions"
Does your system make AI-driven decisions that need an audit trail?
Decision logging records every AI decision with a tamper-evident entry — who made it, what input produced it, when it happened. Required for EU AI Act high-risk systems, financial model risk, and enterprise AI governance.
"I need to verify an artifact or receipt"
Do you have an artifact hash, certificate ID, or receipt ID to check?
Public verification requires no account. Provide the certificate ID or receipt ID and the API returns the full verification record — hash match, signature validity, signing key ID, and all provenance fields.
"I'm documenting AI system components"
Do you need a machine-readable inventory of your AI system's components?
An AIBOM (AI Bill of Materials) enumerates every model, dataset, and dependency in an AI system. CertifiedData can certify AIBOM entries and link them to artifact certificates — giving auditors a complete component graph.
"I need a public audit log"
Do stakeholders or regulators need to see platform-level event history?
The transparency log is a platform-wide, append-only event record. Every certificate issuance, receipt, and policy event is anchored here. Auditors can verify the log independently without access to your internal systems.
"My AI produces outputs that need to be verified"
Do you need to prove an AI output is authentic and unmodified?
Certified Outputs issues a certificate for any AI-produced artifact — generated reports, model outputs, dataset exports. The certificate hash proves the output hasn't been tampered with since certification.
"I'm building for EU AI Act or enterprise compliance"
Does your use case require regulatory documentation of AI system behavior?
CertifiedData's governance stack covers the data-to-decision chain: certify training data, log decisions, anchor events in the transparency log, and expose machine-readable verification URLs that auditors and regulators can check independently.
How to think about the stack
CertifiedData is four composable layers. Use only the layers your workflow requires.
Verification layer
Public APIs — no account needed. Anyone with a certificate ID or receipt ID can verify independently.
Artifact layer
Certify datasets, model outputs, and AI artifacts. Each certificate is an Ed25519-signed record with a permanent hash.
Commerce layer
Policy-gated payments for AI agents. Every transaction produces a signed receipt cryptographically bound to the artifact.
Governance layer
Decision logging, audit trails, and compliance documentation for EU AI Act, ISO 42001, and enterprise governance frameworks.
Key principle
Every layer produces a cryptographic artifact with a public verification URL. The layers compose: a payment receipt can reference a certificate ID, which references a dataset hash — creating a tamper-evident chain from raw data to spend to proof.
Recommended starting paths
Common workflow patterns — follow the steps in order for the fastest path to a working integration.
Agent buying data or compute
Regulated AI system (EU AI Act)
Common questions
Do I need to use all the components?
No. Each component is independent. Certify datasets without payments, use Agent Commerce without certification, or log decisions without either. Start with only what your workflow requires.
What's the difference between synthetic data certification and artifact certification?
Synthetic data certification is scoped to datasets generated by algorithms like CTGAN. Artifact certification is broader — it certifies any AI artifact: model weights, output files, inference results, training datasets. Both produce the same Ed25519-signed certificate structure.
When do I need Agent Commerce vs just a certificate?
If money or value changes hands — an agent is buying data, paying for compute, or executing any transaction — you need Agent Commerce. If you only need to prove an artifact was generated or is authentic, certification alone is sufficient.
Can I verify without an account?
Yes. The verification APIs are fully public. Provide a certificate ID, receipt ID, or artifact hash and the API returns the full verification record — no authentication required.
How does compliance logging relate to the rest of the platform?
Decision logging records AI decisions in a tamper-evident ledger. The transparency log anchors platform-wide events. These are independent of certification and payments — you can use them alone for EU AI Act Article 12 and 19 logging requirements, or combine them with the rest of the stack for a full provenance chain.
Explore all surfaces
How it works →
Platform overview — the full Generate → Certify → Verify → Log chain.
Synthetic data certification →
Certificate structure for synthetically generated datasets.
AI artifact certification →
Certify any AI artifact with Ed25519-signed proof-of-provenance.
Certified outputs →
Artifact → payment → receipt → public verification.
Agent Commerce →
Policy-gated payments for AI agents. Full lifecycle.
Policy engine →
Set spend limits, currency rules, and merchant allowlists.
Spend governance →
Per-agent and per-policy spend controls with daily/monthly caps.
Receipt schema →
Full payment_receipt.v1 field reference.
Payment verification →
Public receipt verification — hash check + Ed25519 signature.
Agent Commerce use cases →
All use case patterns: marketplace, dataset purchases, and more.
Dataset purchase flow →
Create → attachLinks → capture with certificate_id binding.
Agent marketplace →
Multi-agent marketplace payment and certification patterns.
Synthetic data supply chain →
Full certify → sell → deliver → verify pipeline.
Verify a certificate →
Public certificate verification — no account needed.
Artifact registry →
Browse certified artifacts in the public registry.
Transparency log →
Platform-wide event log for audit.
AI decision logging →
Tamper-evident decision records for AI systems.
AI governance →
Enterprise AI governance and compliance.
Governance framework →
Framework for auditable, compliant AI systems.
AIBOM →
AI Bill of Materials — component inventory for AI systems.
EU AI Act compliance →
How CertifiedData addresses EU AI Act requirements.
AI risk management →
Risk management documentation for AI systems.
Ready to start?
Free sandbox. No credit card. A signed, verifiable artifact on your first request.