SYNTONIC

About Syntonic

The problem: the pairing is unmeasured

AI is in the workflow. Whether it's helping is unmeasured. Across 106 studies, human-AI combinations on average performed worse than the best of either alone (Nature Human Behaviour, 2024). Experienced developers were 19% slower with AI in a randomized trial (METR, 2025). The pairing (not the model) is where the value appears or evaporates. And nobody instruments the pairing.

The insight: measure the coupling

Everyone instruments the model. Nobody instruments the working relationship. We treat the human-and-AI as one unit of analysis and score the relationship itself: is the pair on goal, is meaning actually transferring, are errors getting caught. It's an instrument with a stated optimization target and a published protocol stack. Not a vibe.

What we're building

One instrument, three mechanisms:

  • LOCI: a score for the relationship, not the output. Validation in progress.
  • ASE: the evaluator the model can't game: scored from outside the evaluated system's prompt, on a different model family. Robustness experiment in progress.
  • CEED: errors the pairing surfaces that neither party catches alone. Design-partner pilots open.

How each one is doing, honestly staged: the research roadmap.

The toolbelt (a plain-language picture)

The simplest honest description of Syntonic: one founder holding a belt of patent-pending instruments for measuring (and improving) how humans and AI work together.

There are only three things you can ever do with an instrument like that: have it wielded for you, rent it, or buy it.

  • Wielded for you. Available now. We deploy the instruments on your work ourselves. That's Work with us.
  • Rented. Not yet. The APIs and self-serve product are in validation, not yet available. They're on the roadmap, and statuses change when the work lands.
  • Bought. A future conversation. Licensing isn't on offer today.

One door is open. The other two are labeled honestly. We'd rather the picture be accurate than impressive.

The team

Nick Abbate, founder. 15+ years in product and engineering, most recently six years at the payer–provider seam of U.S. healthcare, building HIPAA-compliant insurance APIs and scaling provider-facing SaaS from 6K to 17K practices at Zocdoc. Molecular and cell biology at Stanford; New York since 2010.

The operating model, stated plainly: a founder working coupled with an AI-native system (we call it Lumen) that executes alongside him. Nick's time goes to relationships, scoping, and senior oversight; the system does the building. No inflated team page: the company is the founder, the system, and the instruments. We measure our own coupling the way we propose to measure yours.

Patent + research posture

Patent pending (nonprovisional submitted for filing, June 2026). We don't publish claim details.

The rule that governs every page on this site: if a number on this site isn't sourced or pre-registered, it doesn't ship. Sources are linked where figures appear; protocols are stated before results exist. The research roadmap is the live record.