Portfolio

Portfolio

Explore our products and research — from AI-powered organizational frameworks to enterprise verification, public transport ticketing, and AI-based architecture research.

NEW • PROTOTYPE

cadenza

Provable data integrity for critical energy infrastructure — prove operational data is authentic and unchanged — even against the operator, and without a blockchain.

Outcome: a reading can be shown authentic and unchanged, including against the operator, without a blockchain.

Ed25519 · Merkle provenance · RFC 3161 anchoring · Java 21 · Spring Boot · Go · PostgreSQL · React

NEW • PROTOTYPE

Aurora ERP

Enterprise resource planning for one operator, with a separate ledger per company. Finance, people, and time on a server you run. Not a shared multi-customer service.

Outcome: one operator, a separate ledger per company, on a server you run.

Java 21 · Spring Boot · PostgreSQL · React · Docker · JWT · Ed25519 (wallet auth) · Flyway

DE PRODUCT • LIVE

Leitungstreu

Provable 15-minute Istwerte for Redispatch and flexibility — beside SCADA and the Energy Management System (EMS). Cadenza is the fabric; Leitungstreu is the German product line.

Outcome: a quarter-hour Istwert stays checkable beside SCADA and the EMS. Neither is replaced.

de.grid.telemetry.v1 · Ed25519 · Merkle · RFC 3161 · Java 21 · Go · PostgreSQL · React

NEW • PROTOTYPE

Puma

Event-driven etiCORE eTicket for public transport — buy, activate, and validate tickets, with fleet and control-centre operations in one system.

Outcome: issue, activate, and validate a fare in one event-driven system.

Java · Spring Boot · PostgreSQL · React · RabbitMQ (event-driven path) · Docker · VDV / etiCORE-oriented

RESEARCH SERIES

AI-Based Architecture in Practice

A comparative field study of autonomous agents used as architecture, not as a tool added afterwards. The paper follows one pattern — an agent at a defined operational bottleneck — through a software company, a US mid-size bank, and a healthcare network, and sets out the five-layer model those cases share.

Comparative field study · 5 layers · 5 levels · 3 cases · KPI framework

Buy the book on Amazon

RESEARCH SERIES

AI on Both Sides of the Pull Request

A model that keeps a human in the loop on both sides of the pull request: a generator instance that drafts the change from the specification, and a reviewer instance that checks it against a standards catalogue. The developer confirms the intent before the generator starts and decides the findings before the merge.

Two instances · Human in the loop · Standards catalogue · Revision cycle