Commerce checkout
Point of sale
Catches: A stacked coupon, price drift, or an AI agent's discount pushes an order below the cost-basis margin floor.
Verdict: BLOCK before the checkout token is issued — signed dossier attached.
The Point of Execution
Start where the money moves. Then govern every point of execution.
Decionis owns one thing decisively today: margin protection at the point of sale. But the gate that blocks a bad checkout is not a commerce feature — it is a deterministic decision layer for any high-stakes action. Checkout is live today. The platform is the point of execution: vendor deals, visa applications, code deploys — anywhere an agent or a person is about to commit something you cannot cleanly undo.
Live today
Where the money moves — checkout
Cost-basis margin floors enforced inline at the point of sale, with p95 decision compute under 120ms. Proven on Shopify with 100+ installs and thousands of governed gates.
The expansion
Every point of execution — any workflow
The same signed, deterministic verdict — proceed, hold, block, or escalate — in front of any decision where authority moves and mistakes are expensive.
Post-facto tools record it. The gate decides it.
Most teams already own logs, approvals, and dashboards. None of them sit at the moment of execution — they explain the damage after it lands.
Audit logs
Reconstruct what happened after the fact.
A signed Decision Dossier is created at the execution boundary.
Manual approvals
Prove someone clicked yes — not that the action matched policy.
One deterministic allow / hold / block / escalate verdict before the action runs.
Dashboards & alerts
Page someone after the money already moved.
The out-of-policy action waits; only real exceptions reach a human.
You do not need to be on Shopify or an ERP
Commerce and ERP are the surfaces where we proved it, not the requirement. The Universal Action Gate is a single API/webhook call: any agent, service, or workflow that asks the gate before it acts is governed by the same policy graph and produces the same signed Decision Dossier. No native integration required — just a call at the moment of execution.
Agent-error templates beyond checkout
The same gate catches the expensive mistake in each domain. Checkout is live today; the rest are deployable policy patterns on the Universal Action Gate — the pattern is ready, not a pre-packaged one-click template yet.
Live
Code deploy / infra change
Point of deploy
Catches: An agent or CI action ships a config change that violates a hard policy — disabling auth, opening a security group, or flipping a production flag.
Verdict: BLOCK or ESCALATE before the change applies, with the violated rule cited.
Vendor deals & procurement
Point of commitment
Catches: An agent or rep commits spend or contract terms beyond policy before a vendor reaches production.
Verdict: ESCALATE to the named approver with reason codes; restrain the commitment.
Visa & eligibility submissions
Point of submission
Catches: A managed workflow would submit an application that fails eligibility or completeness policy.
Verdict: HOLD or ESCALATE before submission so an ineligible filing never leaves the door.
One policy graph across many workflows
Point-of-execution governance is not one gate at a time. A Decision Chain links related verdicts across domains into a single accountable record, so a later action can depend on an earlier decision instead of trusting a scattered trail of tickets and logs.
Step 1
Vendor onboarding
Supplier qualification decision — signed, with policy version and evidence.
Step 2
Purchase-order approval
PO checked against spend policy and the qualified-vendor decision above.
Step 3
Payment release
Disbursement gated on both prior verdicts — no orphaned approval, no silent bypass.
Every step is its own signed Decision Dossier; the chain links them into one Execution Ledger record for cross-domain accountability.
Open, queryable Execution Ledger API
RoadmapToday the Execution Ledger is the durable, signed record behind every verdict. The next step — on our roadmap, not shipped yet — is to expose it as an open, queryable API that AI workflows can consume directly: ask the ledger whether a decision was made, which policy governed it, and how authority moved, so cross-domain agents share one accountable source of truth instead of re-deriving trust per system. We will label it live here the day it ships, and not before.
How do I control what executes, at the moment it executes?
Control that arrives after execution is reporting. Execution control means the check runs inside the path the action takes — at checkout, at the payment release, at the tool call — so the outcome can still be changed rather than merely recorded.
6 patterns answer this
Safe code deployment
A pipeline — or an agent with repo access — ships to production, force-pushes, or runs a migration.
1 policy pack
Treasury payment release
A vendor payment, payout batch, or cross-border transfer is released.
3 policy packs
Margin floor at checkout
An order reaches checkout carrying stacked discounts or a drifted automated price.
2 policy packs
Prevent multi-channel overselling
The final unit of an item is sold on two channels within the same sync window.
1 policy pack
Prevent destructive production changes
A command, script, or agent drops a table, deletes a bucket, or tears down infrastructure.
1 policy pack
Cloud spend increase
An autoscale event, a new cluster, or a provisioning script commits ongoing cloud spend.
Its gate and signals
From this term to a governed action
- Safe code deploymentThe recurring action behind this term.
- DevOps & CI policy packThe pack whose rules decide it, rule by rule.
- Try to break it in the sandboxA real signed verdict, no account.
- Deploy on your surfaceAvailable on 3 platforms today.
- Verify the proof yourselfCheck the signature without trusting us.
- Start NowMeasure on your traffic before enforcing.