Skip to main content

AI Conveyor Architecture

Purpose​

This document describes the architecture of the system that supports AI Conveyor: components for managing the portfolio, tracking initiatives, reporting, and integrating with external systems.

Key ideas​

  • Initiative registry — the central component: storing initiative metadata and statuses, and linking them to stages and stage gates.
  • Reporting and analytics layer — portfolio dashboards, KPIs, stage gate and impact reports (see analytics-layer).
  • Integrations — connections to project management systems, data catalogs, and (optionally) ML platforms for tracking experiments and models.

How it works​

The architecture supports the operating model of the AI office: entering and updating initiatives, conducting stage gates, and computing portfolio metrics and economic impact. It can be implemented using corporate tools (Jira, ServiceNow, Power BI, etc.) or a dedicated platform; the data model defines the minimum required structure (see data-model, initiative-entity).

flowchart TB
    subgraph Intake and process
        A[Requests / Owners] --> B[Initiative registry]
        B --> C[Stage-Gate process]
        C --> B
    end
    subgraph Reporting
        B --> D[Analytics layer]
        D --> E[Dashboards / KPIs]
    end
    subgraph External systems
        B -.-> F[Projects / ML platforms]
    end