Skip to content

ConnectSoft AI Software Factory

Target

This page describes intended behaviour or structure. No implementation is claimed unless a section is explicitly marked Current implementation or Evidence.

Product framing (decision D1, 2026-10-08)

ConnectSoft AI Factory is an organizational AI operating and control platform; software engineering and generated SaaS are its first solution pack. This page was written for the earlier "autonomous SaaS factory" framing and describes that pack's vision. Read "the Factory generates SaaS" as "the software-engineering pack generates SaaS on the platform". Scale, autonomy and timeline statements are aspiration, not evidence; technology statements follow the ADR review notes. See the plane map for the current structure.

North Star

ConnectSoft AI Software Factory is a cloud-native, multi-tenant, event-driven AI production platform where specialized agents use governed templates, reusable libraries, knowledge graphs, and GitOps automation to design, generate, validate, deploy, observe, and continuously evolve complete SaaS systems.

This section describes the final-state target architecture of the factory: not a code generator and not a chatbot, but an AI-native software production operating system that turns business intent into running, observable, continuously improving SaaS platforms.

Operating principle

Everything reusable becomes a module. Everything produced becomes an artifact. Everything meaningful emits an event. Everything learned enters the knowledge graph.

What the factory is

The factory is composed of twelve cooperating platforms, organized into planes. Humans direct and govern through the Experience plane; the Control Plane orchestrates; the Agent Mesh does the work; the Knowledge Platform grounds every decision; delivery and runtime platforms ship and operate the output; and governance, observability, and integration cut across all of them.

flowchart TB
    subgraph experience [Experience Plane]
        Studio["Factory Studio"]
    end
    subgraph control [Control Plane]
        CP["Control Plane<br/>orchestration, policy, state"]
    end
    subgraph work [Work Plane]
        Mesh["Agent Mesh"]
        Knowledge["Knowledge Platform"]
        Templates["Template &amp; Library Platform"]
    end
    subgraph delivery [Delivery & Runtime Plane]
        DevOps["DevOps / GitOps"]
        Runtime["Runtime Cloud"]
        Generated["Generated SaaS Products"]
    end
    subgraph cross [Cross-Cutting Plane]
        Obs["Observability &amp; Feedback"]
        Gov["Governance / Security / Compliance"]
        Integration["Integration Platform"]
        Marketplace["Marketplace"]
    end

    Studio --> CP
    CP --> Mesh
    Mesh --> Knowledge
    Mesh --> Templates
    Mesh --> DevOps
    DevOps --> Runtime
    Runtime --> Generated
    Runtime --> Obs
    Obs --> Knowledge
    Gov -.governs.-> CP
    Gov -.governs.-> Mesh
    Integration -.connects.-> CP
    Marketplace -.publishes.-> Templates
Hold "Alt" / "Option" to enable pan & zoom

The twelve platforms

Platform Role
Factory Studio Human command center: design, review, monitoring, governance.
Control Plane Orchestration, project lifecycle, policy, routing, state.
Agent Mesh Autonomous specialized agents that plan, design, code, test, deploy.
Knowledge Platform Knowledge graph, vector memory, artifact store, context builder.
Template & Library Platform Template-first, agent-enriched generation machinery.
Marketplace Ecosystem catalog of reusable factory assets.
DevOps / GitOps Bridge from validated artifacts to running systems.
Runtime Cloud Execution environment for generated SaaS.
Observability & Feedback Trust and improvement loop.
Governance / Security / Compliance Policy and trust layer.
Integration Platform External tool and vendor integration.
Generated SaaS Product Platform The output: complete running SaaS products.

See the Ecosystem Map for the full picture and the Platform Capability Map for capability-to-platform mapping.

The production lifecycle

The factory operates as a continuous software production loop. Every stage is traceable, replayable, and observable.

Prompt / Business Intent
→ Blueprint
→ Knowledge-Enriched Agent Workflow
→ Validated Artifacts
→ GitOps Delivery
→ Running SaaS Runtime
→ Observability Feedback
→ Knowledge Graph Improvement
→ Better Future Generation

The full diagram and walkthrough are in the Final Reference Flow.

How it stays trustworthy

  • Traceability — a single traceId links intent → blueprint → artifacts → commits → deployment → telemetry. See the Metadata Schema.
  • Reusability — templates, libraries, and patterns are first-class, versioned, and discoverable.
  • Autonomy with guardrails — agents act autonomously but every output is validated and every sensitive action passes a governance gate.
  • Observability-first — logs, metrics, traces, and feedback are designed in, not bolted on.
  • Multi-tenant scale — tenant isolation and horizontal scale are built into every platform.

Technology grounding

The factory and the SaaS it generates are built on the real ConnectSoft stack: .NET 10 / ASP.NET Core, Clean Architecture and DDD, MassTransit on Azure Service Bus, Microsoft Agent Framework for agent runtime, Blazor MFE for experiences, NHibernate persistence on Azure SQL/PostgreSQL, Qdrant vector memory, Serilog + OpenTelemetry + Application Insights observability, Azure DevOps GitOps, and Pulumi IaC. See Architecture Principles and the Technology Stack.