In August 2025, VALESKA begins as a working repository and a software design document. The immediate purpose is practical: discover and index documents across local machines, network storage, and remote systems without moving the original files; classify them through the APQC process framework; and make them available through semantic search.
This is the point where an architecture becomes software. The first design establishes a file scanner, metadata extraction, document classification, embeddings, a PostgreSQL and pgvector foundation, a semantic search service, and planned Alfresco integration. It also treats permissions as part of the work. A useful result is not simply a document that matches a query; it must remain connected to its source, owner, location, and access rules.
The new repository does not begin from a blank page. It gives a contemporary technical form to an idea that had been developing for years: enterprise knowledge should follow the work that creates value.
Knowledge must follow the work that creates value
In 2004, while working at Northrop Grumman, I encountered value-stream thinking in manufacturing. A value stream makes visible the connected activities that turn an input into something useful for a customer. It exposes handoffs, waits, rework, responsibilities, and the places where a process loses momentum.
I began to wonder why enterprise content management was normally organized in a completely different way. Documents were usually arranged by department, by an abstract subject taxonomy, or by the habits of the people who created folders. Those structures might be familiar, but they did not describe how the organization actually produced value.
In a manufacturing setting, a value stream connects work across functions. A customer need may pass through engineering, sourcing, production, quality, delivery, and support. The work is connected even when the reporting hierarchy is not. I believed the same principle could apply to enterprise knowledge: the information used to do the work should be organized around the process and value stream that produced it.
That was the beginning of the Value Stream Knowledge Architecture, which is now known as VALESKA.
From architecture to an executable system
The idea was not simply to redraw a folder tree. A process-centered architecture changes what information means inside an organization. A document becomes an artifact of work: an input, a decision, a handoff, a control, a reusable method, or evidence of an outcome. Its position should help a person understand what it is for, who uses it, and what comes before and after it.
I presented the concept at Northrop Grumman as a way to connect value-stream analysis and enterprise content management. It asked for a substantial change in how people thought about information, ownership, and structure. At the time, the proposal was too radical to become the organization’s operating model. The idea, however, was not abandoned. I continued developing it independently.
Early VSKA material documented the approach through value-stream architecture, controlled vocabulary, repeatable process structures, and security aligned to the people responsible for the work. The question was always how to make the architecture operational rather than leave it as a diagram.
As the architecture took shape, I wrote Architexturizer in C++. Its purpose was to move the work beyond a paper model and toward an executable system. The name reflected the task: give an architecture enough structure (texture) that it could be constructed, applied, and used.
Why the repository matters now
The first VALESKA repository makes several of those long-standing requirements concrete. It establishes non-destructive discovery, so the system can learn about a distributed body of records without claiming ownership of every original file. It establishes metadata extraction, so a record has identity and context. It uses APQC classification to connect information to a common process framework. And it adopts semantic embeddings so people can search by meaning as well as literal terminology.
The planned integration with Alfresco is equally important. A knowledge architecture has to work within the systems where organizations already manage records, permissions, and daily operations. VALESKA should make that environment more intelligible and more discoverable, not require an organization to abandon its existing information landscape.
There is still substantial work ahead. Semantic retrieval must be tested against real collections. Classifications must be reviewable. Permissions must remain enforceable. The architecture must show that it can support many sources and preserve the relationship between a result and the work that produced it.
The beginning of a documented evolution
This repository inception marks a new chapter, not the invention of the idea. The originating insight came from value-stream work; the early VSKA material developed its process, vocabulary, and security model; Architexturizer explored how to make the architecture executable. The 2025 repository begins the next phase: a governed, searchable implementation.
Future entries in this series will be dated to the primary evidence behind them—an original document, a repository commit, or a HECL decision record—and written from the knowledge available at that point in the evolution.
Historical basis: VALESKA initial repository commit be8a50a, August 8, 2025; VALESKA_SDD.md. The Northrop and Architexturizer material is David Gargan’s firsthand account, with early VSKA documents retained as supporting artifacts.