Solutions / AI agent platforms

You run agents for thousands of customers.
Give each of them provable accountability.

Embed AIR in your platform and give every customer signed, anchored proof of what their agents did, and why. Shipped as your feature, under your brand.

EU AI ActNIST AI RMFISO 42001SOC 2GDPR

Talk about embedding See the evidence ↓

White-label · multi-tenant · your customers verify it themselves

White-label capsule · live

Tenantacme-corp

Agentsupport-resolver-12

Powered byAIR · embedded

Actionrefund issued · in policy

PII exportblocked · SV-EXFIL

BLAKE3 · Ed25519 · per-tenant chain · anchored Rekor

✓ your customer verifies — zero calls to you

01The stakes

last 30 days · platform: your-agents-cloud

Your platform ran 4.2M agent actions for your customers last month.

When their auditor or regulator asks what an agent did, and why, that's your platform's problem, not theirs. Most platforms can replay what happened; almost none can show why.

02The standards

Agents are their own regulated category now.

The frameworks

the landscape

Agent governance spans the EU AI Act, NIST AI RMF, ISO 42001, SOC 2 and GDPR — none designed for agentic AI.

NIST opened an AI Agent Standards Initiative (Feb 2026); Singapore published the first agent governance framework (Jan 2026).

Execution + intent

what's usually missing

A defensible agent trail needs both execution (what the agent did) and intent (why). Most programs capture only the first.

AIR's causal graph and counterfactual replay are the "why" — the half most tools can't show.

EU AI Act

Articles 12 & 73

Logging (Art. 12) and incident reporting (Art. 73) are the agent-relevant obligations. The Digital Omnibus (provisional agreement May 7, 2026) defers high-risk Annex III obligations from Aug 2, 2026 to Dec 2, 2027.

Provisional until adopted and published in the Official Journal (expected before Aug 2, 2026); until then Aug 2, 2026 stays an active date. Article 5 prohibitions have applied since Feb 2025.

03How AIR answers

Everything an OEM needs, as an embeddable layer.

Give every customer their own evidence?

Multi-tenantPer-tenant signed chains, cryptographically isolated, one customer never sees another.

Prove not just what, but why?

CausalCausal graph plus counterfactual replay — intent observability, not just execution.

Map one record to whatever framework they cite?

ReportOne AgDR record → SOC 2 · ISO 42001 · EU AI Act · NIST evidence.

Ship it as your own feature?

EmbedThe MIT airsdk under your brand; "powered by AIR" is optional.

Survive agent-to-agent handoffs?

HandoffCross-agent chain of custody — the AgDR Handoff Protocol carries proof across delegations.

04The evidence

Your brand on the record. Your customer verifies it.

Each customer gets a signed, anchored, per-tenant chain they can verify independently, with zero calls to your backend and zero to ours. You ship the accountability; they own the proof.

Read the SDK docs →

AgDR record · agdr/v2 · tenant: acme-corp

Delegation · acme-admin authorized support-resolver-12

Action · refund issued $240 — within policy

Why · causal graph: ticket #88 → policy match → approve

Halt · PII export blocked — SV-EXFIL

sig: ed25519 · hash: blake3 · per-tenant · anchor: Rekor

✓ verified by the customer · zero calls to your platform

05What you get

The tiers platform builders embed.

Enterprise / OEM

embed & white-label

Multi-tenant, per-customer signed chains
White-label, "powered by AIR" optional
Causal graph, query & counterfactual replay
AgDR Handoff for agent-to-agent custody
SOC 2 · ISO 42001 · EU AI Act · NIST evidence

Talk about embedding

Air-gapped

for your regulated tenants

Everything in Enterprise, plus

On-prem / offline anchoring per tenant
No phone-home — records never leave
Extended retention + PQ re-anchoring
Admissibility Pack — FRE 902 + expert support

Talk to us

The OSS AIR SDK is on PyPI today, MIT-licensed: start embedding in an afternoon, then turn on multi-tenant chains, replay and framework reports when your customers ask.

Hand your users proof, not promises.

Let's scope an embed: per-tenant chains, the causal "why," and the framework reports your customers' auditors ask for, shipped as your feature.

Talk about embedding →