The Logic Layer · Ultra Premium

SOMA-Class AISelf Organizing Meta Architecture

Most AI automations run tasks. SOMA-class AI understands your business. SOMA adds a logic layer that checks the AI's work against your business rules before it acts — so the output holds up instead of quietly drifting. This is the tier nobody else can offer.

Ontology Deployment

Your business logic formalized into a machine-readable knowledge graph. The AI doesn't guess — it operates within the boundaries your rules define.

Admissibility Engineering

Every AI action is checked against formal rules before execution — so out-of-bounds actions get caught before they run, not patched after.

Automated Reasoning

Logic programming over your processes. The system derives conclusions, detects anomalies, and improves itself over time without manual intervention.

Self-Correction

When the environment changes, the system recognizes the change and adapts its behavior. Not a chatbot that hallucinates — a system that knows what it knows.

We build it with you — or you don't pay for the setup.

SOMA-Class is a hands-on, high-touch engagement. Isaac personally bootstraps the ontology for your business over a quarter: we encode your rules and stand up the check layer with you. If we don't get that layer standing, you don't pay for the setup. This isn't a product you install — it's intelligence infrastructure we build together.

Talk to Isaac about SOMA

A quarterly cohort — only a few seats each quarter.

← Back to AI Automations
Receipts, not testimonials

Don't take my word for it — take the code.

The logic layer is real, and the work under it is public: the engine is open source, the demo is a real site, and the notes are out in the open.

Open source

The engine, on GitHub

Installable and public: heaven-framework, chaincompiler, carton-mcp, skilltree. Read it, run it, fork it.

github.com/sancovp →
Live demo

A real site we built

Dickinson Plumbing — a working, converting site produced by the system. Not a mockup.

See the build →
Open notes

The research, in the open

The frameworks and notes behind the work — published as we go, not locked in a deck.

Read the notes →