Research

Questions weship answers to.

We are a lab first. Every program below started as a question and ended as something you can hold, download or deploy.

In plain English

These are the big questions our lab works on — like how to make AI cheaper and greener, how to keep it private, and how to make it remember what matters. Each one turned into something you can use: an app, a device, or a system we now build for clients.
SL-R00Active

Power-aware expert routing

“How much energy does AI waste answering easy questions with giant models?”

SPLII treats power as a first-class signal: it simulates specialist-first and generalist plans before running either, under eco, balanced or performance modes. Our architecture projects a 40–80% compute reduction per request; per-route energy metering is the next milestone.

See the work
SL-R01In production

Model routing & minimum context

“Can a router pick a better model than a human — and send it less data while doing it?”

SPLII.ai routes every request across 111 frontier and lightweight models, and passes only the slice of memory the task needs. We study routing quality, cost, latency and privacy together.

See the work
SL-R02Live

Portable, user-owned memory

“What does memory look like when it belongs to the person, not the model?”

The synnc.ai Brain links people, promises, projects and decisions into a living graph that any MCP-compatible AI can read — with validity windows, supersession and a source excerpt behind every answer.

See the work
SL-R03Prototype

Perception at the edge

“How much intelligence can live inside a camera the size of a deck of cards?”

With Pivion we run OCR and recognition fully on-device, turning live video into structured data without a cloud round trip — then hand control to an iPad.

See the work
SL-R04In development

Passive health sensing

“Can care improve when nobody has to tap a screen?”

VitalIQ explores continuous, screenless collection of heart rate, HRV, SpO₂, temperature, sleep and mobility, summarized by specialist health AI for staff and families.

See the work
SL-R05Live

Vision-language products

“What happens when a phone camera can explain what it sees — and what it’s worth?”

Our consumer apps put multimodal models in front of hundreds of thousands of people: valuing objects, detecting scams in screenshots, and scanning rooms for hidden devices.

See the work
SL-R06Active

Search in the age of answers

“How do brands get cited when AI writes the answer?”

We research how generative engines select and cite sources, and turn it into a repeatable playbook for structured data, content and technical SEO.

See the work

Research that never leaves the lab is a hobby. Ours ships — to the App Store, to factory floors, to care facilities, and into the products we build for you.

Live system · SL-R01

SPLII · privacy proxyreq_7f3a92
Where it came from
IP 73.42.118.9iPhone 16 · iOS 26Account u_48213Long Island, NY

“Summarize what I promised the Acme team last week”

SPLII · privacy proxy

forwards req_7f3a92 · question + shared context · from Synnc servers

Anthropic
OpenAI
Gemini
Grok
Mistral

✓ Providers see Synnc

Every call leaves from our servers under our credentials — your IP, device and account never reach the provider.

✓ Request IDs ≠ user IDs

Each call carries a fresh ID, not your account ID.

✓ You choose the context

The provider receives what you type and any memory you connect — nothing else.

✓ Encrypted in transit

TLS from your device to Synnc, and from Synnc to the provider.

Let’s build the impossible thing.

Hardware, software, AI or all three. Tell us what you want to exist — we’ll tell you how fast it can.

Start a project
hello@synnclabs.comText or call us at any time · (934) 300-1042