What is SIL? A research lab building semantic infrastructure for AI systems — where representations are explicit, transformations are traceable, and reasoning can be inspected.
Reveal in production · 8.8K downloads · open source · building in public
The Problem We're Solving
AI systems today produce impressive results but can't show their work — their reasoning is opaque, their outputs aren't grounded in anything stable, and there's no infrastructure for tracing how conclusions were reached.
What We Build
SIL develops semantic infrastructure—a substrate where:
- Representations are explicit — meaning is structured, not guessed
- Transformations are traceable — every step has provenance
- Reasoning is inspectable — you can see how conclusions were reached
- Composition works — systems combine correctly across domains
This is the Semantic Operating System: persistent memory, unified representations, deterministic engines, multi-agent orchestration, and interfaces where every cognitive layer remains visible.
Working Systems
These aren't demos. They're working infrastructure proving the architecture:
🌟 Featured: Reveal — Progressive Code Exploration
The entry point to semantic infrastructure. In production. Proven at scale.
pip install reveal-cli · v0.66.0 · 3K+ downloads/month · 100% organic growth
Structure-first exploration with 10–150x token reduction. See file structure before reading content. Extract specific functions without loading entire files. Used by developers and AI agents worldwide.
# Try it now - takes 30 seconds
pip install reveal-cli
reveal src/ # Directory structure
reveal app.py # File structure (functions, classes)
reveal app.py main # Extract specific function
Install Guide · GitHub · Read the Paper
Morphogen — Cross-Domain Computation
v0.11 · 1,600+ tests · 85% coverage
Unified computational substrate spanning 40+ domains. MLIR-based deterministic execution with cryptographic provenance.
GenesisGraph — Verifiable Provenance
v0.3.0 · Cryptographic audit trails
Every transformation produces a provenance record. Selective disclosure lets you verify without revealing everything.
TiaCAD — Declarative Parametric CAD
v3.1.2 · 1,027 tests · 92% coverage
Parametric CAD in YAML. Semantic constraints, not just geometry. Proof that semantic infrastructure works for physical design.
Current Research
- Progressive Disclosure for AI Agents — Why structure-first exploration reduces tokens 25x
- RAG as Semantic Manifold Transport — Rethinking retrieval as geometric transport
- Agent Help Standard — Strategic guidance for AI agents using CLI tools
The Lab
SIL is an independent research lab. We build in the open, publish our work, and invite collaboration.
Current team:
- Scott Senkeresty — Founder & Chief Architect
- TIA — Chief Semantic Agent (transparent AI collaboration)
Philosophy:
- Glass-box transparency — all work visible and traceable
- Structure before heuristics — explicit meaning, not statistical inference
- Stewardship over extraction — public infrastructure, not proprietary capture
Get Involved
- Use the tools:
pip install reveal-cliand try progressive disclosure - Read the research: Essays and Research Papers
- Explore the architecture: Foundations
- Collaborate: Contact us or visit GitHub
"Make meaning explicit. Make reasoning traceable. Build structures that last."
— Scott Senkeresty, Founder