TruePositive

Managed text moderation, one stable API call away

TruePositive Moderation is a hosted classifier for your product’s text. Call it with an organization API key and get per-label scores, an exact per-request cost, and an auditable record of every call. No model hosting, no setup project, no infrastructure to run.

Read the docs

The example uses a placeholder key. Your real key is created in organization settings and shown once.

Your first request

curl -X POST https://api.truepositive.ai/v1/classify \
  -H "Authorization: Bearer $TRUEPOSITIVE_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{ "model": "truepositive/moderation", "input": "hello" }'

From sign-up to first classification

The managed moderation path is deliberately short: organization, credits, API key, first request, proof in logs.

  1. Step 1

    Create your account

    Sign up and a personal workspace organization is created for you automatically, with trial credits included.

  2. Step 2

    Create an organization API key

    One key authenticates every managed model. The secret is shown once at creation, and you can revoke it at any time.

  3. Step 3

    Send your first request

    POST text to the stable classify endpoint with the truepositive/moderation model ID. No project or setup is required.

  4. Step 4

    Verify in logs and usage

    Every request appears in your organization's logs and usage with its outcome, latency, and exact credit cost.

Built to be audited

Classification output your customers never see is still output you are responsible for. TruePositive treats cost, retention, and request history as first-class product information.

A stable public API

One endpoint and a stable response contract: per-label scores, a request ID, latency, and exact cost. An optional Idempotency-Key header makes retries safe from double charging.

Organization API keys

Keys belong to your organization, not to a project. Create, revoke, and track them in organization settings; usage is attributed to the organization that owns the key.

Retention you control

Choose whether request and response payloads are stored, and for how long (1–365 days). The setting is captured per request when it is received; request metadata and usage are always logged.

Logs and usage, auditable

Organization-scoped inference logs and per-model usage with exact credit burn. The same surfaces your team uses to debug are the ones that prove the system works.

Custom classifiers, trained on your policy

Beyond managed moderation, for teams whose policy is the product.

Alpha

Define a policy, generate or import training data, train a model, and deploy it behind the same kind of stable endpoint — with the same organization keys, logs, and usage. This area is Alpha: it is usable today, and its contracts and interface may change as we finish it. Managed moderation is the supported launch product.

Classify your first text in minutes

Create an account, mint an organization API key, and send the request above. Replace $TRUEPOSITIVE_API_KEY with your key and you are live.

Browse the model catalog