Quickstart

Get TrustPlane running inside your VPC

30 minutes to a governed, multi‑model request path with SLOs‑as‑code and policy‑driven routing.

  1. 1) Install SDK
    # npm
npm i @TrustPlane/sdk
# pnpm
pnpm add @TrustPlane/sdk
# yarn
yarn add @TrustPlane/sdk
  2. 2) Declare an SLO Manifest

    Define latency, reliability, budget, and safety as code. TrustPlane enforces this at the request path.

    # TrustPlane.slo.yaml
slo:
  latency_p50: 300ms
  reliability: "99.9%"
  budget_monthly: "$45k"
  safety: pii_detect+mask
policy:
  allow_models: [gpt-4o, claude-3-5-sonnet, llama-3.1-70b]
  regions: [aws-us-east-1, gcp-us-central1, azure-eastus]
routing:
  objective: cost_aware_latency
  failover: automatic
  canary: true
  3. 3) Initialize the client and make your first call
    // app/api/example/route.ts
import { NextResponse } from 'next/server';
import { createClient } from '@TrustPlane/sdk';

const bp = createClient({
  sloManifestPath: 'TrustPlane.slo.yaml'
});

export async function POST() {
  const resp = await bp.chat({
    model: 'auto',         // TrustPlane picks model + region + hardware
    input: 'Summarize Q2 risk reports for the board',
    team: 'risk-ops',
    trace: true
  });

  return NextResponse.json(resp);
}
  4. 4) Observe cost & policy adherence

    Every request emits latency, availability, and cost per team. Policy hits and PII handling are auditable.

    • Cost per request & per team
    • Policy hits, redactions, and guardrails
    • SLO adherence (latency, reliability)
    • Exportable audit logs
  5. 5) Deploy in your cloud(s)

    TrustPlane runs in your VPC (AWS/GCP/Azure). Use marketplace or Terraform modules to roll out the gateway/sidecar.