Provenance
Every number traces back to the system that produced it and the rule that interpreted it.
Each layer has a precise responsibility and talks to the next through versioned APIs and events. Systems remain the source of truth: Artiplane reads them, reorganises them and makes them operable by AI.


Architectural principle
The first layer is the only one that touches source systems. Upper layers work on shared entities, processes and policies; every value keeps its link to the source and every write returns to systems only through authorised actions.
Read-only connectors via API, CDC, files and events. Bridge documents applications, data, code and screens and publishes objects with lineage, permissions and freshness.
The same entities held in ERP, PLM, POS and CRM are reconciled into one model (product, variant, season, stock, customer). Exposes Entity API and a semantic index.
Versioned BPMN processes with roles, thresholds and approval gates. Every task knows which systems it uses, which data it reads and who decides.
Breaks objectives into tasks and coordinates AI agents, LLMs, n8n workflows and systems with queues, retries, idempotency and human control.
Role-based operational views combining several systems: priorities, evidence and actions ready to approve, with outcomes recorded in audit.
Bridge connects read-only to authorized systems and reconstructs how the company really works: applications, data, code, screens and processes. The result is a living corporate second brain that AI and people can query.
Approved scope, least-privilege credentials, no writes.
Application map, data dictionary, code rules and user flows.
Owners confirm gaps and answers before publication.
Validated knowledge feeds Process Engine, Orchestrators and Super Apps.

SAP, AS/400, WMS and custom applications stay where they are. Bridge makes them understandable and usable by AI, extending their useful life.

The Process Engine turns knowledge into executable BPMN processes: every step knows which systems it uses, which data it reads, who decides and where AI can help.
Lanes, gateways and human approvals on pre-mapped Fashion models.
Systems, data, roles and AI agents for each step.
Time, cost and criticality to see where to act.
Proposals only on manual, automatable tasks.
Every action is authorized, traced and consistent with business processes.
Architecture and securityCross-cutting guarantees
Every number traces back to the system that produced it and the rule that interpreted it.
No action reaches a system without permission, policy and, where required, human approval.
Inside the product
Select a layer to see it in the console.

Real product screens · Atelier Nova Group reference tenant
From signal to action
Example: a line’s sell-through drops in three stores. Each layer adds a piece; none duplicates data.
Bridge reads POS sales and ERP stock where they already live.
SKU, size, colour and store are reconciled on the fashion model.
The Process Engine triggers the reallocation process and its owner.
The Orchestrator computes the proposal with the authorised LLM and thresholds.
The planner sees the proposal in the Super App and approves it.
Every layer described on this page can be navigated in the demo.