NexArt

What is NexArt?

NexArt creates tamper-evident evidence of what an AI system ran. A lightweight SDK records inputs, parameters, outputs, and context, then seals them into a Certified Execution Record. An attestation node signs each record, and anyone can verify it later without an account, helping teams answer auditors, regulators, and customer disputes.

Top Features:

• Certified records: each AI run becomes a signed, timestamped artifact that resists tampering.
• Confidential mode: hash commitments replace raw prompts and outputs to keep data private.
• Public verification: third parties check a record's integrity with no login or API key.

Use Cases:

• Audit requests: hand auditors portable evidence tied to the original AI execution.
• Decision disputes: show which inputs led to a lending or insurance decision.
• Agent workflows: record tool calls and actions across a multi-step agent session.

Who Can Use NexArt?

• AI engineers: add evidence capture with one SDK call and no infrastructure changes.
• Compliance teams: prepare for EU AI Act, SOC 2, and NIST reviews.
• Lenders and insurers: defend customer-facing AI decisions on eligibility, claims, and pricing.

Pricing

• Free ($0): 100 certified runs monthly, two API keys, and full SDK access.
• Pro ($6,000 per year): 5,000 certified runs monthly, five API keys, and email support.
• Team ($18,000 per year): 50,000 certified runs monthly, multiple environments, and priority queue.
• Enterprise (from $50,000 yearly): dedicated attestation node, custom limits, SLAs, and audit retention.

Pros and Cons

Pros:

• Independent proof: verification never depends on trusting the company running the AI.
• Free local use: creating and verifying records costs nothing, only certification is paid.
• Easy integrations: works with LangChain and n8n, plus its own SDK and CLI.

Cons:

• Pricey upgrade: the jump from free to $6,000 a year is large.
• Technical setup: developers must add the SDK to existing AI pipelines.
• Not observability: it proves executions but does not replace logs or traces.

FAQs:

1) What is a Certified Execution Record?

A sealed, signed record of an AI run's inputs, outputs, parameters, and context.

2) Do verifiers need an account?

No, anyone with the record can check its integrity using the public key.

3) Can sensitive data stay private?

Yes, confidential mode swaps raw inputs and outputs for cryptographic commitments.

4) Does it replace logging tools?

No, it complements observability by adding portable proof where logs fall short.

5) Is there a free plan?

Yes, it includes 100 certified runs monthly with a hard cap.