Integrations

Connect the systems you already run

Open standards, a documented API, and a no-code agent workflow you build yourself. Most systems connect with configuration rather than a project.

Browse the catalogRequest an integration

The shape of it

Content and identity flow in; the platform reaches back out

How VIDIZMO connects to the systems you already run Systems on the left connect to the VIDIZMO platform in the centre. Content and identity flow inbound through ingestion connectors, SSO and SCIM. The platform reaches back out through its REST API, webhooks, embedded widgets and apps. Beneath, the agentic layer reaches any MCP server or REST endpoint you define. SYSTEMS YOU ALREADY RUN Identity providerEntra ID, Okta, Ping Case, records, business systemsTyler, Karpel, SAP, Epic Cameras and VMSAny RTSP or ONVIF source Storage and contentS3, Azure Blob, SharePoint SSO · SCIM INGESTION · RTSP VIDIZMO platform One access model, one audit trail, one retention policy Deploys in your cloud, your datacentre, or fully air-gapped Live queries run as the asking user, so your system still decides what comes back API · WEBHOOKS WIDGETS · APPS REACHING BACK OUT Your applicationREST API and widgets Your event handlerWebhooks on what happens Your SIEMTelemetry over OTLP Your archiveExport on a stored query THE AGENTIC LAYER, WITHOUT WAITING ON OUR ROADMAP Bring an MCP serverany system publishing one is reachable with configuration alone Describe a REST endpointit becomes a step in an agent workflow

How another system reaches VIDIZMO

Four ways in, all shipping today

Every path below is available to every deployment, including on-premises and air-gapped ones.

REST API

The full platform surface: upload, search, case and evidence management, metadata, users, access rights and sharing, documented with an OpenAPI specification.

Calls authenticate as an identity and resolve through the same permission model as the interface, so the API cannot exceed what its caller could do directly.

Webhooks

The platform calls your endpoint when something happens, so an integration reacts to events instead of polling: media actions, metadata changes, live sessions and workflow completion.

Delivery is logged, and the event catalogue tells you what you can subscribe to.

JavaScript widgets

Put platform functionality inside your own application: players for video, audio, image, document and playlist, library views, search, transcription, maps and quizzes.

An embed resolves against the viewer's identity, so access control travels with it.

App model

Integrations are apps, enabled per portal and typed by what they do: identity, provisioning, tracking, support, AI processing, conferencing, content management, storage, import, export, partner, MFA and LTI.

Without waiting on our roadmap

Two ways to connect a system yourself

Neither requires us to build anything for you.

Bring an MCP server

If a system publishes a Model Context Protocol server, AI Intelligence Hub connects to it as a client with configuration alone: a name, a URL, a transport and a header for the token.

Tools are discovered from the server at run time rather than coded in, so a server that gains a tool is usable without reconfiguration. Streamable HTTP, Server-Sent Events and WebSocket are carried. VIDIZMO consumes MCP servers; it does not publish itself as one.

Describe a REST endpoint

Point the HTTP request node at an endpoint you define, with the authentication and response handling it needs, and it becomes a step in an agent workflow.

The same node is available as a tool the model calls on its own judgement, so an agent can decide when to reach your system rather than following a fixed path.

No-code, low-code, and open standards

An unlisted system is usually configuration, not a project

The fastest way to add an integration is to build it yourself, without a developer. AI Intelligence Hub carries a no-code visual workflow designer: drop an HTTP request node onto the canvas, describe your endpoint and how it authenticates, and that system is reachable from an agent workflow. Point the MCP node at a Model Context Protocol server instead and its tools are discovered at run time, with no mapping to write at all.

That is low-code where you want control and no-code where you do not, and neither waits on our roadmap or on an engineering ticket from us.

Underneath it, much of what looks like an integration is simply a standard we implement, which is why a system we have not named by brand is usually still reachable:

  • SAML 2.0, OAuth 2.0, OpenID Connect — any compliant identity provider, configured per portal.
  • SCIM 2.0 — accounts created, updated and deactivated from your directory.
  • RTSP and ONVIF — any IP camera or VMS that re-streams. No new hardware, no camera vendor lock-in.
  • LTI 1.3 and SCORM — any conformant LMS, with deep linking and grade passback.
  • OpenTelemetry (OTLP) — traces, metrics and logs to any compatible collector or SIEM.
  • Model Context Protocol — any system publishing an MCP server.

Before procurement asks

What a security reviewer usually wants to know

Does an integration become a way around access control?

No. When an agent queries a connected system live, the query runs on behalf of the person asking, so the source system applies its own rules and decides what comes back. Content ingested into the index carries the permissions it was ingested with.

What works when the deployment is disconnected?

Outbound paths do: the REST API, ingestion and export connectors, telemetry export to a collector inside your own boundary, and self-hosted models that make no external calls.

Anything depending on an inbound call from a vendor's cloud needs a reachable endpoint, so air-gapped deployments should plan the integration around that constraint.