RDFL — Federated Learning
IEEE ICDISM 2023 · NTU Singapore
Manufacturers share yield intelligence without exposing proprietary process data. Accuracy within 1–2% of centralized training — IP stays on-prem.
Suppliers share intelligence, not data.
Federated learning, blockchain provenance, and proprietary material-process-performance models — production-grade from day one.
Data → Intelligence → Transactions → Financing, on shared foundations of connectors, federation and security — every layer engineered for privacy-preservation, auditability and production-grade scale.
Trade finance, invoice discounting and ESG-linked credit underwritten on live production data.
Verified procurement, RFQs, sales orders, digital material passports and blockchain-anchored provenance.
Material-process-performance models, defect attribution, yield simulation and supplier scoring — with the MatiE copilot on top.
Unified MES, LIMS, ERP, supplier and IoT data streamed into one lot-level material ledger.
MES · LIMS · ERP · supplier portals · IoT line telemetry · TMS/WMS — auditable from sensor to settlement.
RDFL federated learning trains across manufacturers without moving raw data. BLIC anchors passports, transactions and signatures on-chain.
Zero-trust networking, customer-managed keys, SOC2-aligned controls and cryptographic provenance for IP-sensitive manufacturers.
Train across customer datasets without moving raw data — each manufacturer benefits from collective intelligence while keeping IP private.
Anchored material provenance and transaction records — immutable audit trail for CBAM, SBTi, and customer disclosures.
Material-process-performance models built with NTU & SERIS. The only dataset linking lab-grade material properties to GW-scale production yield.
Every transaction strengthens the supplier graph. Every onboarded line refines the AI. Switching cost compounds with every datapoint.
Three peer-reviewed programs power data intelligence, supply-chain trust, and yield optimization — built with NTU Singapore and validated in production.
IEEE ICDISM 2023 · NTU Singapore
Manufacturers share yield intelligence without exposing proprietary process data. Accuracy within 1–2% of centralized training — IP stays on-prem.
Suppliers share intelligence, not data.
IEEE Cybernetics 2019 · NTU Singapore
Immutable records and cryptographic authentication for every material transaction and supplier qualification — the trust spine of the network.
Supplier trust is cryptographically verifiable, not self-reported.
NTU Singapore · IHPC A*STAR
AI-enabled blockchain for transparent solar PV supply chain management with ML optimization and end-to-end traceability — validated with Tata Power Solar.
Proprietary solar dataset built over years of live deployments — cannot be purchased or scraped.
Triple-layer IP advantage. No competitor combines privacy-preserving collaborative AI (RDFL), blockchain-authenticated provenance (BLIC), and a live solar PV intelligence platform (KONARK) in one system. Every new manufacturer deepens the moat.
Two decades of accumulated knowledge in advanced materials for solar, batteries, and semiconductors. Nuances of material behavior competitors cannot replicate.
Patent-pending models trained on real-world manufacturing data — improving continuously as more manufacturers join. A flywheel effect that widens the lead.
Curated ecosystem of pre-qualified suppliers with proven track records. Network effects make the platform more valuable as it grows.
Built-in carbon accounting and compliance reporting for CBAM, SBTi, and customer disclosures — a critical differentiator as regulations tighten.
Deep understanding of clean-tech manufacturing from decades of hands-on experience. We speak the operator's language — not a generic SaaS pitch.
"AAGM doesn't just provide data — they understand our manufacturing challenges and speak our language. That expertise is irreplaceable."
Join the operators, suppliers and partners building the operating system for the global clean-tech economy.