Home model and CANVAS
Rooms, floors, areas, devices, capability-aware controls, dashboards with revision history, and household users with distinct roles.
HELIX is in private alpha. This page describes the current product posture without treating a planned capability as a shipped one.
Each item below has been verified running on the HELIX test appliance, not only in the codebase. Last audited October 12, 2026.
Rooms, floors, areas, devices, capability-aware controls, dashboards with revision history, and household users with distinct roles.
Each integration runs as its own process over a versioned MQTT protocol, so one failing vendor does not take down the home.
Ask about the home, control supported devices, author dashboards and automations, and get explanations grounded in recorded events.
Sunrise and sunset with offsets, schedules, room presence, device changes, spoken phrases, and school-day or weekend conditions, running every day in a real home.
Sleep, waking, comings and goings, and room use learned from sensors, calendars, and sleep records, with the evidence behind every statement and a way to correct it.
A dry run that compares the automations HELIX would write against the ones that exist: keep, refine, add, or retire, each with its reasoning.
Each sensor is judged by its own reporting rhythm; a device that goes quiet is asked to report again and, if it does not, logged in a repair ledger the household can see and close.
Models are checked at start-up and routed by capability; an unavailable or failing model is replaced rather than silently degrading answers.
Install, link, enable, disable, restart, and remove integrations, with credentials kept separate from configuration.
Containerised add-ons for ESPHome, Zigbee2MQTT, a Matter bridge, and an OpenThread border router, managed from HELIX.
Matter devices commissioned onto HELIX's own Thread network, including smart locks.
Google calendars sync into HELIX; school calendars, days off, and custody schedules tell it who is expected home, and when.
Changes HELIX proposes, from automations to device actions, wait for a household member to approve or decline them.
HELIX remembers household facts and the corrections people make, and uses them in later answers and decisions.
One physical device reached through several integrations, such as a TV over two connectors, is treated as one thing.
Alarm clocks set in conversation or on the dashboard, rung on the home's speakers.
Each of these has been connected and used in the HELIX test home. Trademarks belong to their owners; listing them does not imply endorsement.
Read-only scan, curation, preview, commit, and rollback are built and tested; not yet proven on a full migration of a real home. Importing Home Assistant automations works today.
Drafting a connector from a vendor's API description, validating it, and refining it against the live API has worked in testing; installing generated connectors is still being finished.
HELIX proposes automations and alerts from observed patterns; the review, learning, and household-control experience is still being refined.
Voice satellites, speaker recognition, and an adjustable assistant personality work; appliance-grade, whole-home voice is still in progress.
The integrations above work; broad parity across the residential device market is not claimed.
Scheduled weekly backups of the database, identity, settings, and credentials run today; restoring a home from a backup has not yet been exercised on a real home.
Electricity tariffs and per-device power profiles are recorded; household energy insight is still being built.
Room-scoped displays and installable dashboard widgets exist and are being tested; they are not yet a finished experience.
Turnkey hardware, guided setup, recovery, support, mobile polish, and compatibility expected by mainstream households.
Conflict detection, health scoring, simulation, explainability, and safer change recommendations.
Conversational layout generation, direct manipulation, per-device layouts, revision history, export, and templates.
Continued expansion across device families and local-first protocol paths based on vendor documentation and field demand.
The product story should evolve only when the implementation, test evidence, and real-device behavior justify the claim.
HELIX needs advanced households willing to work through actual devices, difficult integrations, network constraints, and long-lived automation behavior.