Patent Mapping
Unsterwerx is a document-domain implementation of concepts from US Patent 9,069,626 (Trusted Client-Centric Application Architecture) by Dr. Robert Whetsel. It covers normalization, governance, lifecycle management within the document domain. Broader cross-application control described in the patent remains a long-term direction.
This page maps every major patent concept to its current status in the codebase.
Concept mapping
| Patent concept | Status | Unsterwerx realization | Code evidence |
|---|---|---|---|
| Native applications / heterogeneous inputs | Implemented (narrowed to documents) | Local files plus import adapters for ChatGPT, Notion, Obsidian, Telegram exports | src/import/adapters/, src/parse/ |
| Normalized Application Containers (NACs) | Implemented | Per-format parsers (PDF, DOCX, XLSX, PPTX, TXT, CSV, Markdown) and import adapters that normalize native data into the Universal Data Set | src/parse/*.rs, src/import/adapters/*.rs |
| Universal Data Set (UDS) | Implemented | Canonical markdown plus structural elements, stored in content-addressed storage, indexed via FTS5 | src/canonical/store.rs, src/canonical/search.rs, src/canonical/mod.rs |
| Universal Data Module (UDM) | Implemented | SQLite database + CAS filesystem for canonical content and diffs + provenance tracking | src/storage/, migrations/*.sql, Context::init in src/lib.rs |
| Shared Sandbox | Implemented | Local data directory (~/.unsterwerx/) containing the database, canonical store, diffs, archive, templates | src/lib.rs (Context::init), src/core/config.rs, src/storage/ |
| Business Intelligence | Implemented | Classification rules, source hierarchy weights, Bayesian knowledge scoring, BI deduplication | src/rules/hierarchy.rs, src/bayes/, src/knowledge/, src/semantic/ |
| User Intelligence | Implemented | Retention policies with scope cascade, legal hold, mutability flags, archive actions | src/rules/engagement.rs, src/rules/policy.rs |
| Denormalization | Implemented | Reconstruction from canonical store to markdown or read-only PDF using Tera templates | src/reconstruct/ |
| Synchronization integrity | Partial | Append-only hash-chained audit log, pipeline run IDs, dedup rollback, provenance tracking | src/audit/logger.rs, src/audit/query.rs, src/core/pipeline_run.rs, src/import/provenance.rs |
| Universal Interface | Partial | CLI command surface covering all operations; the patent envisions a richer cross-application workspace | src/cli/ |
| Universal Controller | Not yet implemented | The patent describes a distributed orchestration layer for coordinating many applications; current pipeline orchestration covers a subset | No single controller module |
| Identity / permission management | Not yet implemented | The patent describes cross-domain credential brokerage and permission management; current code has governance scopes and trust weights but no multi-user auth subsystem | Scope metadata in src/rules/policy.rs, src/rules/engagement.rs; no auth module |
| Connection pool / live transport | Not yet implemented | The patent describes a runtime transport fabric between applications; no equivalent exists | None; no module in src/ |
| Bidirectional sync into target applications | Not yet implemented | Import is one-way into the sandbox; reconstruction outputs markdown and PDF rather than writing back into native applications | src/import/, src/reconstruct/ |
Where the implementation is strongest
Normalization into a common intermediate form. This is the core architectural move in the patent. Seven format-specific NAC parsers convert native documents into canonical markdown stored in content-addressed storage. Every downstream operation (search, similarity, diff, classification, reconstruction) works from that normalized representation, not the originals.
Policy-aware operation in a trusted local environment. Not just a label: local-only storage by default, hash-chain audit verification, provenance tracking, scoped policy resolution, signed-PDF immutability, rollback-aware BI deduplication. Source comments tie specific behaviors to patent claims (e.g., src/rules/policy.rs references Claims 5-6 for cascade validation).
User-defined business and lifecycle logic. Classification rules, source hierarchy weighting, Bayesian scoring with user feedback, scoped retention policies, legal hold, dedup ordering with rollback. This goes beyond what a generic file parser would offer.
Where the implementation diverges
Not a general application-integration runtime. The patent describes a runtime that sits between live applications, accepts input through a universal interface, and pushes denormalized data back into target systems. In practice, Unsterwerx is a document normalization and policy-governed lifecycle system. It processes document exports, not live application streams.
Identity and cross-domain sharing are absent. Governance scopes and trust weights exist in the codebase, but not the cross-application credential brokerage or sharing fabric described in the patent.
CLI as Universal Interface. A reasonable first-domain choice. The patent envisions a richer end-user workspace spanning multiple applications; the CLI covers document operations only.
Related board items
Several board items track work that extends patent coverage:
- #36 Scoped governance enforcement (advancing User Intelligence)
- #59 Patent positioning language (ensuring accurate framing)
- #29 Bayesian knowledge scoring (advancing Business Intelligence) -- done