ES
← Back to Portfolio
Agentic AI & Automation June 2026 Proprietary

Mantia, Anonymized Hub of Agentic Tools for SAP Plant Maintenance

An anonymized hub cataloguing a family of agentic tools oriented to SAP Plant Maintenance (PM). Each tool automates a maintenance workflow (work-order generation, spares planning, notification triage, scheduling and more) on one shared agent core with a mock-or-real connector to the enterprise system. Delivered under confidentiality: general description only, no client data.

Composition
Portal hub + 3 specialist agents
Shared core
OData V2 (+CSRF) connector + agent base + web baseline
Connector mode
Mock (dev) ↔ real (prod), same agent code
Deployment
Live (private); dual-target VPS + Azure Container Apps
Architecture diagram of Mantia, Anonymized Hub of Agentic Tools for SAP Plant Maintenance
#agentic-ai #maintenance #odata #eam #llm #automation #private

Business Context

Maintenance is where uptime is won or lost, and most of its planning work is structured enough to assist with agents, but only if the assistant speaks the enterprise system's language safely. Mantia's mock-or-real connector lets the agents be developed, demoed, and validated without touching a production backend, then promoted to the real one unchanged.

Strategic Value

Mantia demonstrates a reusable accelerator pattern: a shared agent core (connector + agent base + web baseline) that turns each new use case into a thin, fast-to-build specialist agent. The hub makes the suite legible to a business audience (catalog, lifecycle, business case), and the mock-first connector de-risks integration with heavyweight enterprise systems.

The Challenge

Industrial maintenance planning is repetitive, knowledge-heavy, and bound to enterprise systems whose data is awkward to work with. Each task (raising a work order, planning spare parts, triaging a notification) needs the same enterprise context but is usually done by hand, one transaction at a time.

Our Approach

Mantia is a hub-and-agents suite. A portal hub presents a data-driven catalog of agents with lifecycle tiles and a business case per agent. Three specialist agents sit behind it (one drafts maintenance work orders, one plans spares and reservations against material stock, one triages plant notifications) and all share a single agent core: a generic OData V2 connector (with CSRF), an agent base, and a common web baseline. The connector runs against a faithful mock of the enterprise schema in development and a real backend in production, so the same agent code ships to both.

Key Performance Indicators

KPIBaselineResultImpact
New use case → working agentBespoke build per integrationThin agent on a shared coreFaster, consistent delivery
Integration riskDev against a live backendMock-first connector, real in prodBuild & demo without touching production

Proprietary, source code not publicly available

Architecture

mantia architecture

mantia architecture

A suite, not a one-off

Mantia is a private suite of agentic assistants for industrial maintenance. A portal hub catalogs the agents; each specialist (work-order generation, spares planning, notification triage) is a thin layer on one shared agent core, talking to an enterprise EAM schema through a connector that is a faithful mock in development and the real backend in production.

This is proprietary work; the live deployment is private and behind access control. The card describes the architecture and intent without exposing client data, names, or internal logic.

Technology Stack

Python FastAPI Pydantic-AI LiteLLM HTMX OData Docker Azure Container Apps

Visual assets for this project are not publicly available.