New

Synheart's infrastructure is live on device inference is free, forever

Synheart

Developers · Lab

Structure for real experiments, on top of live human state.

Lab is a Core SDK surface for recording structured experiment data alongside the live HSI stream. Run a real protocol, baseline, intervention, recovery, and capture it as first-class structure, without pausing the state signal your app already depends on.

  • Research-tier
  • On-device by default
  • Runs alongside the live HSI loop, doesn’t interrupt it
Glass and orange layered block, standing for a structured record

Why Lab

Your protocol, as a first-class object.

A normal HSI stream tells you how someone’s state changed. It doesn’t know your protocol has a baseline phase, then a stressor, then a recovery window. Lab adds that structure on top, without changing how the live engine runs underneath.

  • No Separate Recording Pipeline

    Lab runs on the same live engine your app already uses, sample push and HSI emission continue normally.

  • Real Protocol Structure.

    Nest windows (baseline, stressor, recovery) the way your study actually works, not as a flat log.

  • Ground Truth When You Need It

    Assert known values for a specific window when you, the experimenter, know something the signal alone wouldn’t capture.

What you can do

What Lab gives you

Man thinking at a laptop in a busy office
  • Protocol-scoped sessions

    Start a session under a named protocol, like “morning baseline plus five-minute stressor.”

  • Nested windows

    Open and close labeled windows inside a session, baseline, stressor, recovery, however your protocol is structured.

  • Window annotations

    Attach values to windows, mood scores, intervention IDs, and labels. Record a known value for a time slice as fact.

  • Ground-truth overrides

    Assert a known value for a slice of time when you want to record it as fact rather than inferred state.

  • Device and user metadata

    Attach consistent device and user context to every session export automatically.

How a session works

From protocol to export.

  1. Session

    Start a session under your protocol.

  2. Open Window

    Open a window for each phase of your protocol, baseline, stressor, recovery, nesting as needed.

  3. Annotate

    Annotate any window with your own values as the study runs.

  4. Close Windows

    Close windows and finalize the session. The live engine never stops running underneath.

Laptop keyboard with a cable plugged in

Lab is gated by a capability flag. Check availability before assuming the surface is present, on a runtime built without it, Lab methods simply report unavailable rather than failing.

Boxes on a conveyor belt

In the workplace

Research-tier, on purpose.

Lab session data is sensitive, structured research data, so it’s held to a higher bar than the standard HSI stream.

  • Cloud upload of a Lab session requires explicit research consent from the person, on top of standard consent.
  • It also requires a research-tier capability grant on your app, not enabled by default.
  • Without both, Lab sessions stay entirely on-device. Nothing is silently uploaded.
  • Research data is routed and retained separately from standard HSI uploads, under research-tier policy.

Where Lab fits.

How it fits with the rest of Synheart

  • Core SDK

    Runs the live HSI engine that Lab records structure on top of.

  • Lab

    Adds protocol structure, windows, and ground truth on top of the live stream.

  • Platform

    Exported Lab sessions can be reviewed where research capability and consent are set.

  • CLI

    Useful for local testing of Lab flows with mock signal data before running a real session.

Why Synapse

Plans include free on-device inference.

Every plan includes on-device inference at no cost. You only pay for what crosses into the cloud.

Woman smiling at her phone

FAQ

Frequently Asked Questions

Build with a model that knows the moment.

Works alongside your existing Core integration. See full pricing on the Pricing page.