What is Syntonic?

Syntonic builds the coupling-quality layer for human-AI interaction. Lumen is the first instrument built to this architecture — a live demonstration of what becomes possible when coupling fidelity is the design target.

The name comes from syntony — the condition of being tuned to the same frequency, in resonance. In radio engineering, a syntonic circuit receives a specific signal while rejecting noise. We use this precisely: a syntonic system is one where information passing between nodes is accurate, coherent over time, and structurally defended against distortion.

The thesis

Scaling laws have dominated AI for a decade. More parameters, more compute, more data. Syntonic proposes an orthogonal axis: coupling laws. The right measurement is not intelligence-per-parameter but resolution-per-energy — and the evaluator must be architecturally unable to see the system prompt.

This is the layer most AI products quietly skip. Models get bigger; couplings get noisier. Without coupling-quality measurement, "better AI" becomes a measurement gap dressed as progress.

Shared intelligence

Intelligence until now has been isolated. It lives inside a single mind, a single model, a single session — and it stays there. Syntonic is built on a different premise: the most powerful intelligence doesn't live inside either party. It emerges in the space between them. An emergent mind that neither party contains alone. We call this shared intelligence.

Shared intelligence has always existed. It's what happens between two minds that have learned to think together — finishing each other's sentences, anticipating the next move, building something neither could reach alone. But AI changes the equation. For the first time in history, there's a form of non-human intelligence capable of joining that process. Every AI you've ever used is fluent, but it doesn't know you. The longer you use it, the more that gap shows. Syntonic is what closes the gap over time. Shared intelligence that compounds. That holds. That gets smarter every time you come back.

Syntonic doesn't just know what you say. It knows what you mean.

The architecture

The core invention is an evaluator structurally prevented from seeing the generator's system prompt: Goodhart-immunity by architecture, not policy. The evaluator scores coupling quality but cannot be gamed by the generator, because it cannot see the generator's instructions.

Combined with LOCI — the Learning-Optimal Coherence Integral — as the universal coupling-quality metric, and Distributed Instance Coherence for unified identity across surfaces, this forms the coordination substrate multi-agent AI systems will need at scale.

Underneath all of this sits the Canon — a graph-aware, Obsidian-backed memory substrate built from original research at the intersection of information theory, cognitive architecture, and AI alignment. The Canon persists across sessions and anchors every coupling to an accurate, evolving model of the problem space.

Most AI systems are trained to maximize user satisfaction. This produces a predictable failure mode: outputs that are locally coherent and comfortable, but that drift from the actual state of the problem. The drift is structural — it's what optimization for satisfaction looks like at scale. Syntonic's architecture runs counter to this at the generative layer. The only valid output is an accurate one. The difference matters most when the accurate answer is one you didn't want.

The instruments

Lumen Live — voice, avatar, and chat. The flagship coupling instrument.

Lumen Skill + MCP Connector — coupled AI as a first-class citizen inside Claude Desktop, with persistent memory across sessions and surfaces.

Concordance — per-coupling-partner onboarding and calibration.

Canon — the graph-aware, Obsidian-backed memory substrate.

Who built it

Nick Abbate (Stanford '07, B.S. Honors Molecular & Cell Biology) is the founder of Syntonic. He spent eight years at the payer-provider seam of U.S. healthcare — most recently as Principal Product Manager at Zocdoc, where he built HIPAA-compliant insurance APIs and scaled the provider-facing SaaS platform from six thousand to seventeen thousand practices.

The Canon was developed through original research mapping how information fidelity degrades across minds, institutions, and AI systems, and what restores it.

Substrate

Syntonic, Inc. is a Delaware C-corporation, incorporated April 13, 2026. Based in Brooklyn, NY.

The architecture is protected by USPTO Provisional Application #64/038,092 — thirty-eight claims across eight families, priority date April 13, 2026. Counsel: Michael Jones, Jones IPL (Brooklyn).

The trademark SYNTONIC™ is on file with the USPTO under Serial No. 99741989.

First Founder Circle pilots in motion.

This is a preview

Syntonic is currently invite-only. The Canon, the coupling architecture, and the interface are all actively evolving. If you have access, you're part of the founding cohort — and what you notice matters.

Questions or feedback: hello@syntonic.io