ValarIQ

Independent control plane for autonomous action

Control what AI agents are allowed to do.

ValarIQ sits between autonomous systems and execution. Identity, authority, policy and evidence before sensitive actions reach your rails.

Request access See how it works

RefundsPaymentsProcurementTreasuryAccount ops

Design partner operate path

Discover · Agents · Connect · Decide · Prove · Insights

Find what is acting, promote identity, wrap the money path, settle ESCALATE, keep signed evidence — then ask Ops Chat what is going on. Discovered ≠ authorised.

Product overview Capabilities Trust & Security

Same agent. Same capability. Different authority.

Refunds, payments, procurement, treasury and account changes on one control path.

ALLOW

Within delegated authority.

£45 refund · straight through
£120 supplier invoice · within mandate
Payment lookup · read-only

ESCALATE

Human decision required.

£250 refund · named Approver
£8,000 procurement · dual review
KYC address change · sensitive write

DENY

Outside permitted authority.

£1,250 refund · above band
New beneficiary payment · not on allowlist
Treasury transfer · out of hours

Built for teams deploying agents on money paths

Policy in the loop. Not people in every loop.

Agents reason, select tools and initiate money movement without a person on every step. Controls built for humans stall automation. Four questions decide whether that is safe to deploy.

01
Who is this agent?
02
Who authorised it?
03
What is it allowed to do?
04
Can you prove why this happened?

Without ValarIQ

Agent asks. Human checks every financial action. Automation stalls. The benefit of autonomy disappears.

With ValarIQ

Agent asks. ValarIQ evaluates authority. Routine actions proceed. Exceptions reach people. Evidence is captured automatically.

ValarIQ answers all four before execution.

One control point

Observe. Govern. Verify. Control. Prove.

Start with payments, refunds, procurement and finance operations. ValarIQ authorises. Your stack executes.

01
Observe
Know what is acting
Discover agents and AI surfaces, then promote into the registry. Discovered is never the same as authorised.
02
Govern
Define authority
Identity, ownership, mandates and policy. Financially sensitive authority first; same model for the enterprise.
03
Verify
Check the claim
Signed evidence and context assurance on decisions. Historical truth stays fixed when the world moves.
04
Control
Allow. Escalate. Deny.
The Action Gateway evaluates proposed actions before they reach your stack. ValarIQ does not execute.
05
Prove
Leave the record
Attributable decision history within verified scope. Tamper-evident, not absolute immutability theatre.

Synthetic vignette

A £45 refund that stays true when limits move.

Illustrative story only. Not customer data. Same pattern applies to payments, procurement and treasury: historical decisions stay fixed while current reliance can require re-verification.

Decision time

Refund agent requests £45. Within authority. ValarIQ returns ALLOW. Signed evidence seals who, why and what was decided.

Authority changes

Later, the agent’s ceiling is cut. The original ALLOW is not rewritten. Historical truth remains intact.

Current reliance

For a claim today, Evidence Lifecycle flags re-verification. Automate the routine; escalate when the world moved.

See the full product journey →

Try it

Watch three decisions land.

Pick a scenario. See allow, escalate or deny from the same control path. Deploy in observe mode first; enforce when Risk is ready.

Synthetic illustration. Not production monitoring or customer data.

Agent request

action: tools:refund
context: {
  "payment_id": "pay_1042",
  "amount_gbp": 250
}

Select a scenario, then press Evaluate policy. Amount bands are workspace-configurable. Thresholds are not fixed in public demos.

0.0%

Would allow

0.0%

Would escalate

0.0%

Would deny

Example observe distribution. Your dashboard shows your fleet.

Why not build this in-house?

IAM says who can access. ValarIQ decides what money may move.

IAM

Who can access?

ValarIQ

Allow · Escalate · Deny

Your rails

Execute

Fail closed

If ValarIQ is unavailable, high-risk money movement stops. Enterprise resilience patterns are on the roadmap.

Your stack stays

Keep Stripe or your PSP for execution. ValarIQ decides whether the agent may call it.

Evidence that stays honest

Signed evidence on every decision. Lifecycle assessment keeps historical truth fixed when applicability changes.

Proof

See the artefact, not just the claim.

Automate the routine. Escalate the exceptional. Block the prohibited.

Put ValarIQ between your agents and your rails. Start with one financially sensitive workflow.

Request access What is live today