Sign inOpen console
Decision automation platform

Business rules live in configuration, not in deployments.

Your applications send context. Decision Control evaluates your logic and answers with a decision, the actions to take and the reasons why. Every time, and replayable.

Open the consoleRead the docs
POST /api/v1/decisions/transaction-monitoring/executesuccess · 39 ms · DEC-874981
Request
{
  "customer_id": "C-1001",
  "amount": 15000,
  "currency": "USD",
  "country": "DO"
}
Evaluation trace
Above 1,000?matched
15000 > 1000
Customer risk check / Amountmatched
15000 > 10000
Customer risk check / Countrymatched
"DO" in [DO, RU, NG]
Decision
HOLD
SEVERITY HIGH

Transaction above 10,000 going to a high risk country.

createAlertsendWebhook
  1. 01Applicationssend context
  2. 02Decision platformevaluates configuration
  3. 03Rules, data, integrationstables, calculations, calls
  4. 04Decision and actionswith the reasons why
What a decision needs

An engine that knows rules, not your industry.

Inputs, conditions, variables, rules, flows and actions. What they mean is up to whoever configures them.

InputAmount > 10k?HOLDcontinue

Visual flows

Compose conditions, calculations, external calls and nested decisions on a canvas, and see how a case travels through them.

AmountCountryOutcome
> 10000in [DO, RU, NG]HOLD
………

Decision tables

Rules in a grid that analysts can read, validate and evaluate on their own, without waiting for an engineer.

ruleAbove 1,000?
actual15000
expected> 1000
outcomematched

Explainable by default

Every execution returns a trace: which rules fired, against what values, and why the outcome is what it is.

Saved testspassed
Simulationpassed
Backtest on past inputsrunning

Test, simulate, backtest

Run saved tests and replay historical inputs before a change ever reaches production.

Draft→Review→v1 published→Production

Versioned and audited

Drafts, publish requests, immutable versions and promotion between environments, with a full audit trail.

RESTPOST …/execute
npm@decision-control/runtime
MCPagents operate decisions

Runs where you need it

Call the API, embed the runtime with no server, or let an AI agent work with your decisions over MCP.

How it works

From a request to a reasoned decision.

  1. 01

    Model

    Describe inputs and rules in the console, or start from a template.

  2. 02

    Validate

    Run tests, simulations and backtests against real data.

  3. 03

    Publish

    Promote a version across environments, with review.

  4. 04

    Decide

    Your applications call the API and receive an output plus its trace.

For developers

Pure engine. No framework in the way.

The engine takes a graph, an input and ports, and returns an output and a trace. All I/O is injected, which is why it is testable in milliseconds and why any decision can be replayed.

  • TypeScript
  • Python
  • Java
  • .NET
Execute a decision
$ curl -X POST /api/v1/decisions/transaction-monitoring/execute \
  -H "x-api-key: $DECISION_API_KEY" \
  -d '{"customer_id":"C-1001","amount":15000,"country":"DO"}'

{
  "decision": "HOLD",
  "actions": ["createAlert", "sendWebhook"],
  "reasons": [
    { "rule": "Above 1,000?", "outcome": "matched" }
  ]
}

Put your next decision in configuration.

Start from a template and have a decision running in minutes.

Open the consoleRead the docs