Unify Logo Footer.svg
API Manager
Logo
Overview

Overview

Logo

4 mins READ

API Manager is a Platform Tools feature that wraps your automations, AI models, external APIs, and event streams as governed REST APIs — with authentication, policies, rate limits, and monitoring included. It acts as a wrapper layer: any resource you have built becomes a secure, observable API without extra infrastructure.

Overview

The API Manager gives you a managed gateway between your internal work on UnifyApps and the outside world. It wraps whatever you have built — an automation, a model, a third-party API — and exposes it as a governed REST endpoint that callers can reach with standard credentials. Instead of wiring integrations by hand, you expose an endpoint, group it inside a collection, decide who can call it via a client and access profile, and apply policies that enforce authentication, rate limits, and transformations. The platform then handles the public URL, auth validation, traffic rules, and the Insights dashboard automatically.

overview_(1)_-new.png
overview_(1)_-new.png

Building Blocks

Five pieces work together to form a managed API:

  • Collection — a group of related endpoints published under one base path and version (for example, "Orders API v1").

  • Endpoint — a single REST operation (a path + HTTP methods) backed by something you built.

  • Client — an external application or partner that consumes your APIs.

  • Access profile — how a client authenticates and which collections or endpoints it may call.

  • Policy — a reusable rule applied to requests or responses: rate limits, validation, CORS, transformations, and more.

A managed API is the combination: endpoints grouped in a collection, exposed to clients through access profiles, governed by policies.

Publishing an API End to End

The typical path from idea to live API:

  1. Create a collection — set its base path, version, and any collection-level policies.

  2. Add endpoints — for each, set the path, HTTP methods, the backing resource (Callable, Webhook, LLM Model, External API, or Event Stream), and the request and response schemas.

  3. Apply policies — attach rate limits, request validators, or other rules at the collection or endpoint level.

  4. Register a client and create an access profile — pick an authentication method and scope it to the right collections and endpoints.

  5. Share the URL and credentials with the consumer, or export an OpenAPI spec, and watch traffic on the Insights dashboard.

What You Can Expose as an Endpoint

An endpoint can front one of five resource types, and each is invoked differently:

Resource Type

How It Is Invoked

Callable

A synchronous automation: the call runs it and waits for its output.

Webhook

An asynchronous automation: the call is accepted immediately and the automation runs in the background.

LLM Model

A large language model exposed as an API; the call passes the payload and returns the model's response.

External API

An upstream third-party API the platform proxies to, adding your own auth, rate limits, and monitoring.

Event Stream

An event topic; the call publishes the request payload as an event.

Notes

  • Create one client per external consumer so you can grant, monitor, and revoke access independently.

  • Apply policies at the collection level for broad rules, then tighten per endpoint or per access profile.

  • Export an OpenAPI spec from any collection to share a machine-readable contract with consumers.

  • Check the Insights dashboard after publishing to confirm traffic is flowing and to catch early errors.

  • The same endpoint mechanics — auth, policies, monitoring — apply to all five resource types.

FAQs

What is the difference between a client and an access profile?

client is the identity of a consumer — the partner, app, or service you're granting access to. An access profile is how that client actually calls your APIs: it bundles the credentials, the scope (which collections and endpoints), and any per-client policies. A client can have multiple profiles (for example, one for production and one for testing), each with its own credentials and scope.

Does the API Manager work with non-automation resources

Yes. In addition to automations (Callable and Webhook), endpoints can front an LLM Model, an External API (proxying a third-party service), or an Event Stream. The same auth, policy, and monitoring mechanics apply to all five resource types.

Can I run multiple versions of the same API at the same time?

Yes. Each collection carries a version label and an Include version in path toggle. When enabled, the version appears in the URL (for example /v1/orders/), letting v1 and v2 run at different paths simultaneously. Callers on the old version stay on /v1 until they choose to migrate.