Documentation

From your tables to a
governed prediction.

A short tour of how AhinsaAI works in practice: connect data, ask a predictive question, and act on a scored, backtested, audited result.

Quickstart PQL Deployment Gateway API
Quickstart

Three steps to a prediction.

1

Connect your data

Each use case declares the tables and columns it needs. Upload to its schema contract, which AhinsaAI validates and assembles into a temporal relational dataset.

2

Ask a predictive question

Write the task in PQL: the entity, the target, and the horizon. The graph engine handles features, training, and scoring.

3

Act on governed output

Get scored predictions with a held-out backtest, risk drivers, and an audit record. Route, review, or escalate. Every action is logged.

Predictive Query Language

PQL expresses a prediction as a single statement. You name what to predict, the entity to predict it for, and the time horizon.

fraud.pql
# Will this account see a fraudulent payment in the next week?
PREDICT payments.is_fraud
FOR EACH account.id
WITHIN NEXT 7 DAYS
churn.pql
# Which customers are likely to churn in the next 30 days?
PREDICT customers.churned
FOR EACH customers.id
WITHIN NEXT 30 DAYS

The same shape covers regression and ranking targets. Labels are computed point-in-time, so there is no forward leakage.

Deployment

Runs where your data lives.

On-premises

  • Inside your data center
  • No external dependencies
  • Full network control

Private cloud / VPC

  • Your cloud account
  • Connects to your warehouse
  • Single-tenant service

Air-gapped

  • No outbound connectivity
  • Offline model serving
  • For the strictest environments

Always

  • Your data stays in place
  • Hash-chained audit
  • Fail-closed authentication

Gateway API

Every use case is reachable over an authenticated REST gateway. Requests are tenant-scoped by API key, and the same path serves the UI and programmatic callers.

run a use case
# Score a use case on your data
POST /v1/run/payment_fraud
Header X-API-Key: <your-key>

# List available use cases and fetch a schema contract
GET  /v1/usecases
GET  /v1/usecases/payment_fraud/contract

Responses include the scored predictions, the held-out backtest, the risk drivers, and a reference to the audit record. See the benchmarks for how those numbers are produced and reported.

Want a guided walkthrough?

We will connect a sample dataset and run the full path with you: query, score, backtest, and audit.

We will reach out within one business day.