Faceless Agentic ERP and the Rise of Agentic Enterprise Systems
Autonomous AI Agents That Transform Your ERP from a System of Record into a System of Execution
No screens. No forms. No manual workflows.
Celebal Technologies, Faceless Agentic ERP
This white paper presents Celebal Technologies' strategic framework for "Faceless ERP," an architectural paradigm that transforms ERP systems from passive systems of record into active systems of execution. The core problem: large enterprises run 30–45 fragmented ERP instances, creating data silos, process rigidity, and growing dependence on scarce specialist talent. The newer wave of vendor-native AI agents only worsens this, replacing ERP fragmentation with agent fragmentation, since each agent sees only its own data.
The proposed solution keeps ERPs in place but abstracts away their interfaces. Rather than navigating screens, forms, and transaction codes, users interact through natural language while intelligent AI agents reason and act across systems. Celebal's recommended approach is a Lakehouse-centric architecture (built on Databricks), where agents live "where the data lives." Since the Lakehouse already holds 80%+ of needed enterprise data, most workloads run there, while lightweight micro-agents sit at the ERP only for real-time needs like inventory—governed by a "Latency Decision Framework."
The architecture spans six layers: user, orchestration, enterprise agents, semantic/ontology, data platform, and systems of record. Key enablers include an enterprise ontology for consistent metric definitions, Agent Bricks for orchestration, and Lakebase for persistent agent memory. Real-world results are cited: a $10M projected cost reduction in order-to-cash contract automation, 96% precision in manufacturing root-cause analysis, and 6–10× productivity gains for 1,600+ users at a Fortune 500 firm. A governance framework covers design, integration, deployment, observability, and compliance.
