Skip to content
Gatewise

Product

An action control plane between agents and business software

Gatewise is an action control layer for AI agents. It evaluates proposed business actions against deterministic policies and trusted evidence before execution.

Request accessSee how it works

Works with your agents and existing business tools.

Capabilities

What the product does

  • Action gateway

    One entry point for proposed actions. The agent asks before it acts, so there is a single place where control is applied.

  • Evidence checks

    Decisions use current system data: invoice state, contract dates, ownership, consent, prior actions. Missing data returns INSUFFICIENT EVIDENCE.

  • Policy engine

    Deterministic rules with thresholds, hard blocks and conditions per action type. The same action and evidence always give the same decision.

  • Approval routing

    Actions above an autonomous limit are held and routed to a named approver with the full context attached.

  • Duplicate detection

    Repeated or already covered actions return NO ACTION instead of a second message, a second task or a second refund.

  • Decision record

    Every proposed action is stored with the policy evaluated, the evidence used, the decision, the reason and the execution status.

Decisions

Five outcomes your application can rely on

Integration is simple because the contract is small. Your code handles five cases and nothing else.

  • ALLOWEvidence and policy conditions are met. The agent may execute the action.
  • BLOCKA policy rule forbids this action. The agent cannot execute it.
  • REQUIRE APPROVALThe action is valid but above an autonomous threshold. A named human has to approve it.
  • NO ACTIONThe action is unnecessary. A duplicate or an existing task already covers it.
  • INSUFFICIENT EVIDENCERequired system data is missing or stale, so the decision cannot be made safely.
Action log
Today

Interface example. Not Gatewise customer data.

Evidence

Decisions built on system state, not on summaries

Gatewise references system categories rather than a fixed vendor list. Evidence sources are configured per policy.

Evidence sources
by category
  • CRMOwnership, stage, lifecycle, record state
  • SupportTickets, conversations, escalations
  • Payments and billingInvoices, refunds, disputes, ledger state
  • Internal toolsTasks, workflows, entitlements

Categories, not integration claims.

Refund policy
Active
  1. R01Allow refunds up to 100 EUR automatically.
  2. R02Require manager approval from 100 EUR to 500 EUR.
  3. R03Require Finance approval above 500 EUR.
  4. R04Never issue a refund when a chargeback or previous refund already exists.
  5. R05Require invoice status = paid.

Interface example of a configured policy.

Policies

Owned by the team that owns the rule

Policies are stated in business language and enforced deterministically. Gatewise does not ask a model to interpret the rule at decision time.

See the policy engine

Shadow mode

Measure before you enforce

Run Gatewise with enforcement off. See what would have been allowed, blocked or escalated, then enable enforcement per action type.

Shadow mode - last 30 days
enforcement off
actions evaluated
1,284
would have been blocked
42
required approval
11
duplicate actions detected
7

Interface example. These numbers are not Gatewise customer results.

Give your agents freedom without giving them a blank cheque.

Tell us what your agents are allowed to do today. We will review whether Gatewise fits your agent workflow.

Request accessSee how it works

Works with your agents and existing business tools