Introduction & Purpose
Relevant source files
The following files were used as context for generating this wiki page:The Keith Soyka Portfolio serves as a static, evidence-based repository for Keith Soyka's professional work in AI systems building and operations leadership. The project presents an "inspectable" record of software builds, particularly the GestaltView runtime, and defines the service lanes for AI implementation, design, and fractional operations.
This site functions as a public gallery of interactive HTML artifacts, bridging the gap between theoretical AI claims and verified technical outputs. It demonstrates a commitment to "evidence before narrative" by labeling every project with its current status: verified, staged, or planned.
Sources: README.md:1-2, index.html:85-90, work.html:36-39, about.html:72-75
Core Philosophy and System Ethics
The system architecture follows an "operator-led" approach. This philosophy prioritizes the business workflow over the technological tool. The project adheres to four core commitments to ensure transparency and reliability in software delivery.
| Commitment | Description |
|---|---|
| Evidence before narrative | Every claim requires a label (verified, staged, or planned). |
| Process before tool | The smallest system that solves the constraint is the optimal choice. |
| Handoff, not dependency | Useful systems leave the user more capable through documentation. |
| Correction is a feature | Systems must allow users to correct and redirect outputs. |
Sources: about.html:72-82, services.html:36-40
The Operator's Lens
The project utilizes a "systems lens" derived from IMAX projection systems. The operator sees the chain of small decisions that make the final output possible. This perspective translates to AI builds by focusing on failure modes and handoff paths.
Sources: about.html:49-55, index.html:150-155
Project Structure and Navigation
The portfolio organizes content into six primary pages and a specialized showcase directory. The structure facilitates a logical flow from high-level value propositions to deep technical artifacts.
The diagram shows the hierarchical relationship between the main landing pages and the specific system components.
Sources: README.md:5-12, showcase/registry.json:173-200, sitemap.html:35-45
The GestaltView Ecosystem
GestaltView v3 is the flagship continuity infrastructure within this portfolio. It is a user-directed system designed to preserve fragmented human context and translate it into actionable next steps.
Key Components of GestaltView
- Live Runtime: Built using a React/TypeScript client, Vercel serverless API, and Supabase/Postgres data layer.
- Billy Agent: A collaborator agent running on Web, Discord, and Reddit, governed by ten constitutional invariants.
- Corpus Governance: Infrastructure for source-authority hierarchy and additive ingestion where rows are never deleted.
- Render Engine: A pipeline that takes raw source material and synthesizes it into finished artifacts (e.g., Markdown, PDF-ready HTML).
Sources: work.html:43-60, work.html:75-85, showcase/render-engine.html:168-185
Artifact Generation Flow
The Render Engine synthesizes source material based on specific synthesis styles and target formats.
The diagram illustrates the process of generating an artifact from raw source material using the live runtime.
Sources: showcase/render-engine.html:105-120, showcase/render-engine.html:410-440
Service Lanes and Technical Advisory
The project defines three distinct lanes of engagement to address different business constraints. Each lane follows a fixed-scope model.
Service Lane Matrix
| Lane | Service Type | Key Deliverables |
|---|---|---|
| Lane A | Advisory & Build | Workflow maps, risk notes, 90-day roadmaps, working AI agents. |
| Lane B | Design Services | Pitch decks, identity systems, web design, and custom software. |
| Lane C | Operations | P&L ownership, labor planning, and multi-site transition management. |
Sources: services.html:43-160, index.html:112-140
Evaluation Methodology
The project introduces the Portfolio Throughline Test. This eval harness checks if an AI system preserves relationships across tasks. It uses a failure taxonomy including categories like preserved, collapsed, distorted, and invented.
Sources: work.html:90-100
Conclusion
The Keith Soyka Portfolio provides a technical foundation for deploying "inspectable" AI systems. It serves to bridge the gap between high-level operations leadership and production-ready software. By grounding every claim in source files and verified artifacts, the project establishes a standard for transparent, operator-led technology builds.
Sources: index.html:90-95, README.md:25-30