https://github.com/tom-tagmdl/concierge
Home Assistant interaction and orchestration integration for room-aware, deterministic, explainable smart home experiences.
https://github.com/tom-tagmdl/concierge
ambient-computing context-aware-computing deterministic-systems hacs home-assistant human-centered-design interaction-design orchestration smart-home voice-interface
Last synced: 9 days ago
JSON representation
Home Assistant interaction and orchestration integration for room-aware, deterministic, explainable smart home experiences.
- Host: GitHub
- URL: https://github.com/tom-tagmdl/concierge
- Owner: tom-tagmdl
- Created: 2026-06-27T21:49:56.000Z (28 days ago)
- Default Branch: main
- Last Pushed: 2026-07-09T02:26:35.000Z (16 days ago)
- Last Synced: 2026-07-09T04:16:29.773Z (16 days ago)
- Topics: ambient-computing, context-aware-computing, deterministic-systems, hacs, home-assistant, human-centered-design, interaction-design, orchestration, smart-home, voice-interface
- Language: Python
- Homepage: https://tagmdl.com/homes-that-behave-well
- Size: 4.77 MB
- Stars: 0
- Watchers: 0
- Forks: 0
- Open Issues: 15
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README

## How It Works

# 🏠 Concierge
**Concierge is the Household Coordination Engine for the Homes That Behave Well platform.**
Concierge transforms governed household knowledge into deterministic, explainable, room-aware experiences.
Rather than controlling devices directly, Concierge coordinates context, capabilities, experiences, identity, continuity, occupancy, restoration, messaging, productivity, and household awareness into a unified interaction system.
---
# ⚠ Current Release Status
## Concierge V1 (Current Release)
The current release focuses on:
- Room-aware interactions
- Alias-driven execution
- Scene activation
- Signals
- Global context
- Dashboard experiences
- Deterministic execution
## Concierge V2 (Roadmap Vision)
The architecture documented in this repository represents the future Concierge V2 platform.
Coordinator V2 expands Concierge into a complete household coordination engine capable of consuming:
- Vocabulary
- Capabilities
- Experiences
- Continuity
- Affinity
- Restoration
- Occupancy
- Presence
- Messaging
- Household Memory
- Productivity Context
- Provenance
---
# 🌎 Platform Position
Homes That Behave Well separates responsibility across several platform services.
```text
Foundation
What is true?
Asset Intelligence
What matters?
Voice Identity
Who is interacting?
Concierge
What should happen?
```
Concierge does not define truth.
Concierge consumes governed context and determines how the home should respond.
---
# 🧠 Coordinator V2
Coordinator V2 is the runtime heart of Concierge.
Coordinator consumes governed platform knowledge and transforms that knowledge into household behavior.
Coordinator consumes:
- Room Vocabulary
- Capability Projections
- Experiences
- Person Continuity
- Person-Room Affinity
- Experience Restoration
- Occupancy
- Presence
- Messaging Context
- Productivity Context
- Household Provenance
Coordinator does not own governance.
Governance remains in **Homes That Behave Well (HTBW)**.
---
# ✨ What Concierge Does
Concierge bridges the gap between:
- what the home knows
- what the home can do
- what the household needs
- what the household experiences
It provides a deterministic coordination layer that creates predictable household behavior.
---
# 🧠 Context-Aware Interactions
Concierge responds based on:
- room context
- occupancy
- presence
- household state
- experience eligibility
- household priorities
Every interaction is grounded in known context.
No runtime guessing.
No ambiguous device selection.
No uncontrolled discovery.
---
# ⚡ Deterministic Planning and Execution
Concierge is designed around predictable behavior.
Execution follows governed models rather than runtime inference.
Examples:
```text
Close the shades
```
→ Executes the appropriate room capability.
```text
Movie time
```
→ Selects the experience appropriate to the room, occupancy, and context.
```text
What should I know?
```
→ Produces a synthesized household briefing.
---
# 🏠 Room-Aware by Design
Every experience begins with room context.
Concierge understands:
- Rooms
- Merged Rooms
- Composite Rooms
- Floors
- Occupancy Zones
The household interacts with places.
Not devices.
---
# 👤 Person-Aware Experiences
Through Voice Identity and governed household models, Concierge can consume:
- Identity attribution
- Room affinity
- Communication preferences
- Restoration preferences
- Household continuity
This allows the home to adapt while remaining deterministic and explainable.
---
# 🔄 Continuity and Restoration
Concierge can preserve continuity across interactions.
Examples:
- Restore media
- Restore lighting scenes
- Restore work context
- Restore household experiences
Every restoration decision remains explainable.
Questions the platform should always answer:
```text
Why was this restored?
Why wasn't it restored?
Why was restoration suppressed?
```
---
# 🚶 Occupancy and Presence Awareness
Concierge consumes:
- Occupancy state
- Presence state
- Confidence state
- Multi-occupant context
This allows decisions to be informed by:
- Who is present
- Where they are
- Confidence levels
- Household composition
---
# 🌍 Household Awareness
Concierge combines:
- Weather
- Calendar
- Email summaries
- News
- Household signals
- Tasks
- Shopping state
- Productivity context
to answer questions such as:
```text
What should I know?
What happened while I was away?
What still needs attention?
```
---
# 💬 Messaging and Notification Discipline
Concierge follows a calm-by-default philosophy.
Messaging behavior is:
- Deliberate
- Explainable
- Occupant-aware
- Room-aware
- Interruption-aware
Concierge focuses on:
- Useful notifications
- Escalation when appropriate
- Acknowledgement awareness
- Household coordination
---
# 📋 Household Productivity
Concierge can surface governed productivity experiences such as:
- Calendar awareness
- Household briefings
- Task coordination
- Shopping coordination
- Status synthesis
- Knowledge experiences
Calendar and task systems remain systems of record.
Concierge consumes context.
It does not replace provider systems.
---
# 🧠 Household Memory and Explainability
Concierge is designed to explain itself.
Core questions include:
```text
What happened?
Why did it happen?
Why here?
Why now?
Why for this person?
Why didn't it happen?
```
Explainability is a first-class platform concern.
---
# 🧩 Core Concepts
## Vocabulary
Defines how the household refers to spaces and experiences.
## Capabilities
What the household can do.
## Experiences
Meaningful household outcomes.
## Continuity
What was happening previously.
## Affinity
Household preferences and context.
## Restoration
How previous experiences resume.
## Occupancy
Who is present and where.
## Messaging
How the home communicates.
## Provenance
Who did what, when, where, and how.
---
# 🧠 Design Principles
Concierge follows the Homes That Behave Well philosophy:
- Calm by default
- Deterministic behavior
- Explainable decisions
- Local-first execution
- Governed architecture
- Configuration over discovery
- Household-facing outcomes over implementation compatibility
---
# 🚫 What Concierge Is Not
Concierge is not:
- Alexa
- Google Assistant
- A generic LLM wrapper
- A device discovery engine
- A cloud-first orchestration engine
Concierge does not:
- Guess device targets
- Scan devices during execution
- Redefine governance
- Replace systems of record
---
# 🏗 Governance Model
## Homes That Behave Well (HTBW) Owns
- Architecture
- ADRs
- Contracts
- Models
- Governance
- Canonical definitions
## Concierge Owns
- Consumption
- Resolution
- Orchestration
- Planning
- Routing
- Execution behavior
Coordinator V2 consumes governance.
It does not define governance.
---
# 📦 Repository Structure
```text
custom_components/concierge/
coordinator/
consumption/
vocabulary/
capability/
experience/
continuity/
affinity/
restoration/
occupancy/
presence/
messaging/
memory/
productivity/
diagnostics/
explainability/
docs/
architecture/
contracts/
models/
governance/
patterns/
philosophy/
```
---
# 🚀 Development
## Install Dependencies
```bash
pip install -r requirements-dev.txt
```
## Run Tests
```bash
pytest -q
```
---
# ✅ Release Requirements
- Pass hassfest
- Pass HACS validation
- Pass tests
- Pass architecture review
- Pass ownership review
- Pass readiness review
- Publish release
---
# 🏁 Vision
Concierge is not a smart-home controller.
Concierge is the household coordination engine for Homes That Behave Well.
It combines governed household knowledge, identity, context, occupancy, continuity, experiences, restoration, messaging, productivity, memory, and provenance into deterministic household behavior.
A home that understands context.
A home that behaves predictably.
A home that explains itself.
A home that behaves well.
---
# 🧭 Learn More
Explore the platform architecture:
```text
/docs/architecture/
/docs/contracts/
/docs/models/
/docs/governance/
/docs/patterns/
/docs/philosophy/
```
---
**Concierge is part of the Homes That Behave Well platform.**