SPL – AI patterns with verifiable truth scores

The minimal, verifiable language to compose intelligent systems from patterns.

License: Apache-2.0 • Version

SPL is the pattern language for quantum AI, it is the first AI language.

• Verifiable: every execution returns structured output plus a truth score
• Composable: patterns declare contracts and dependencies and can be safely combined
• Layered: from L0 meta-structure to L1 critical building blocks up to complete solutions
• Reality-agnostic: digital, physical, and hybrid systems share the same contracts

That’s it. SPL is a small set of rules that makes AI systems predictable, auditable, and reusable.

Every SPL pattern has exactly three structural sections. These map intelligence, execution, and ethics into one contract.

• Intelligence (contract): Greg’s 4 pillars — Goal, Return Format, Warnings, Context
• Runtime (execution): Print AI maxims — Loop, Inspect, and Print
• Culture (guarantees): Human alignment — success, metrics, compliance, safety

These three structures are defined once at L0 (meta-pattern) and inherited everywhere.

SPL borrows quantum ideas to describe composition behavior using classical computation:

• Superposition: multiple candidate patterns can coexist for a goal
• Contextual collapse: runtime selects the best pattern (or weighted ensemble) for the current context
• Entanglement: shared signals/constraints couple patterns so changes in one update truth/risks in others
• Interference: scoring and validation amplify or suppress contributions during composition

All of this is implemented with probabilities, scores, and dependency graphs — no quantum computers required. The result is adaptive behavior with measurable confidence.

• Cognitive patterns (L2) define general capabilities: value-system, knowledge-representation, experience-acquisition, motivation, expectation, ethical-reasoning
• Cross-reality contracts let the same intelligence act in digital, physical, and hybrid environments
• The three structures guarantee alignment and verifiability at every step
• Generalization comes from composition, not model retraining

SPL is how you specify, measure, and align intelligence universally.

Foundational Patterns (v2.3):

• L0 Meta-Pattern: The foundation all patterns inherit from (patterns/meta-pattern/v2.3/meta-pattern.yaml)
• L1 Critical Patterns: 13 universal building blocks (patterns/critical-patterns/*/v2.3/*.yaml)
• L1C Cognitive Patterns: 6 AGI-enabling patterns (patterns/cognitive-patterns/*/v2.3/*.yaml)
• Pattern Registry: Unified index of all patterns (patterns/pattern-index-v2.3.yaml)

Documentation:

• What is SPL? - Start here
• Critical Patterns Guide
• Cognitive Patterns Guide
• Six Layers Overview
• Versioning Policy

Total: 20 foundational patterns that compose infinitely

• Muscle (compute): model capability and raw power
• Structure (SPL): meta-pattern, layers, and the three structures that make behavior predictable
• Culture (ethics): guarantees that encode values and compliance

Pillars mapping:

• Greg’s 4 (Intelligence): Goal, Return Format, Warnings, Context
• Print AI (Runtime): Loop and Inspect (Print via return)
• Asimov-style (Culture): Human protection, human direction, system preservation (expressed via guarantees)

For deeper material, see the docs linked above. Build patterns. Compose them. Measure truth. That’s SPL.