Private project

pumm.group documents Nova and NQIS.

pumm.group is not a company. It is a private project and documentation site by Andreas Pumm. It combines the long-term Nova vision with concrete NQIS engineering work: local, data-sovereign, auditable and step-by-step reviewable.

not a companyprivate project siteno shop

Why this site exists

Nova and NQIS have evolved through local AI, knowledge organization, grounded chat, dashboards, Auth/RBAC, safety, evidence and governance into a granular capability model. A3.0 read-only is currently prepared as the controlled next step.

The focus is the technical idea, its deliberately defined boundaries and why local auditable AI systems follow a different operating model from pure cloud chatbots.

Focus areas

The project is based on four guiding principles.

Local control

Data, sources, logs and system behavior should remain as local and traceable as possible.

Reviewability

Answers, actions, status data and changes should not only happen, but remain explainable later.

Safety

Automation needs boundaries: roles, approvals, backups, emergency-stop mechanisms and blocked risky default paths.

Long-term view

Nova and NQIS are not treated as a one-time demo, but as a gradually growing local platform idea.

Positioning

No commercial presentation, no exaggerated AI promise.

pumm.group combines a private technical project, a book project and a long-term system idea. It is neither a product shop nor a consulting or cloud service.

Nova is not presented as a conscious entity or AGI. NQIS is a local auditable AI and orchestration platform with clearly described technical boundaries.

Project character: private development and documentation site for Nova, NQIS and the book project.
Clear boundary: no shop, no consulting offer, no publicly operated AI service and no AGI claim.
Site contents

What is documented

Nova

The assistant idea, book context, identity, language, knowledge organization and human responsibility.

Nova

NQIS

The current technical state with v2.2-RC runtime, API, dashboards, grounded/stateful chat, memory/knowledge, operations, Auth/RBAC, audit/evidence and the prepared A3.0 read-only candidate.

NQIS

Technology

Architecture principles behind local AI, reviewable knowledge work, role logic, approvals and rollback.

Technology