Meridian · Delivery and routing layer

The delivery layer between approved work and real destinations.

Meridian routes approved Ridian OS outputs to the places work needs to go — local exports, cloud storage, client folders, email drafts, GitHub, APIs, and knowledge stores — while preserving human approval and traceable delivery records.

Layer 01

Reasoning infrastructure

Where intelligence happens.

  • OpenAI Frontier reasoning models
  • Microsoft Copilot / Azure AI Microsoft's AI infrastructure layer
  • Google Gemini / Vertex AI Google's AI infrastructure layer

These are the models that do the thinking. They are not where results are stored or delivered.

Layer 02

Delivery & collaboration

Where work lands.

  • Microsoft 365 & SharePoint Documents, sites, and team spaces
  • Google Drive Shared files and folders

Output destinations your team already uses. These surfaces receive finished work; they do not generate it.

Layer 03

Private enterprise APIs

Your internal systems of record.

  • Internal systems of record CRMs, ERPs, ticketing, custom services
  • Policy-aware access Calls run under your auth and controls

Meridian calls into your private APIs under your policies so agents act on real data, not a copy of it.

How Meridian decides

One delivery surface, three distinct roles.

Meridian routes each approved step to the correct layer — reasoning goes to a model provider, artifacts go to a delivery destination, and system-of-record reads and writes go through your private APIs. Policies, human approval gates, and authentication are enforced per layer so the right system is always doing the right job and every delivery leaves a traceable record.

Applied in practice

See Meridian inside a real institutional workflow.

Piper is a flagship example of the platform in motion — preservice teacher feedback becoming structured insight, stakeholder communication, and approved improvements, with human review at every decision.

Explore a flagship use case