Voice Assistant Integration for IoT Products: Alexa, Google, Siri

Voice control used to be a separate integration project per ecosystem — an Alexa Skill, a Google Action, an iOS HomeKit accessory profile. In 2026, Matter has changed this for many product categories: ship a Matter device, get all three ecosystems for free. But not every category benefits, and some advanced features still require ecosystem-specific work.

The Matter shortcut (when it applies)

For these device categories, Matter handles voice integration across all three ecosystems with no ecosystem-specific code:

• Lighting (bulbs, switches, dimmers)
• Plugs and outlets
• Sensors (contact, motion, temperature)
• Locks
• Thermostats
• Window coverings (Matter 1.3+)

A Matter-certified product is automatically discoverable by Apple Home, Google Home, and Amazon Alexa. Voice commands route through the ecosystem’s voice service to the device. No custom skill required.

This was not the case before Matter. The Matter shortcut is the single biggest shift in voice-IoT integration since voice assistants existed.

When you still need ecosystem-specific integration

For categories Matter doesn’t yet cover well, or for advanced features within covered categories:

Cameras

Matter camera support exists but is less mature than vendor-specific implementations. Most camera products in 2026 ship:

• Vendor app for primary control and recording
• HomeKit Secure Video (for Apple-aligned customers)
• Alexa-compatible RTSP feed (for Echo Show display)
• Google Assistant streaming (for Nest Hub display)

Robot vacuums

Basic on/off and start/stop work via Matter. Advanced features (room mapping, no-go zones, scheduling per room) typically require:

• Vendor app
• Vendor-specific Alexa skill
• Vendor-specific Google Action

Major appliances

Washers, ovens, dishwashers — Matter support exists but most integration still goes through vendor clouds for advanced features (cycle selection, status monitoring, smart-cooking presets).

For these categories, plan for both Matter (basic) and vendor-cloud (advanced) integration paths.

Custom skills / actions — when worth it

A custom Alexa Skill or Google Action makes sense when:

1. The product category isn’t covered by Matter (e.g. complex industrial-IoT visualisation)
2. You need conversation flows beyond “turn on / turn off” (e.g. “Alexa, ask SolarMonitor what was my peak production today”)
3. The product has a strong brand presence and the skill name carries marketing value

Skills/actions are not free. Each ecosystem has:

• A separate development model (Alexa Lambda functions, Google Conversational Actions)
• A separate certification process (typically 2-4 weeks per submission)
• Ongoing maintenance to handle voice service API changes
• A discovery problem — users have to find and enable the skill, which most don’t

For most consumer products in 2026, Matter-only is better than building three skills. Save the skills for cases where they genuinely add value beyond what Matter provides.

Privacy and the always-on microphone

Voice assistants raise privacy concerns regardless of integration path:

• Apple HomeKit — voice processing increasingly on-device (HomePod processes locally where possible). Apple’s privacy story is the strongest.
• Google Assistant — cloud processing by default, with on-device options for some commands. Privacy story improving but cloud-first by design.
• Amazon Alexa — cloud processing, with controls to delete recordings. Privacy story the weakest of the three.

For privacy-sensitive product positioning, lean Apple HomeKit hardest. For broadest market reach, support all three but be transparent about which voice processing happens where.

For broader privacy context see our smart-home privacy post.

Wake word vs explicit invocation

Two voice interaction models:

Wake-word — “Hey Siri, dim the bedroom lights.” The user-facing experience most people expect. Reaches the device via the ecosystem’s voice service.

Explicit invocation — “Alexa, ask MyBrand to dim the bedroom lights.” Used when the product has a custom skill. Less convenient; users routinely forget the invocation phrase.

For Matter-integrated products: wake-word “just works” — the user says “turn on the kitchen lights” and the ecosystem routes to your device. For custom-skill products: explicit invocation is mandatory; “ask MyBrand” friction is real.

The wake-word advantage is another reason to prefer Matter over custom skills when the product category supports it.

The display-screen ecosystem

A growing slice of voice-assistant integration is on display devices:

• Echo Show (Amazon)
• Nest Hub (Google)
• iPad as Home Hub (Apple)
• Smart fridge displays (Samsung)

Display devices show device state visually alongside voice control. For products with rich state (cameras, dashboards, sensors), targeting display surfaces specifically is a differentiator.

What we typically build

For a smart-home product with voice integration in 2026:

• Matter-certified hardware for the core functionality
• Voice integration testing across Apple Home, Google Home, and Alexa during certification
• Vendor-app for advanced features the Matter cluster set doesn’t cover
• Optional custom skill only if the product justifies the maintenance cost
• Display-aware UX for products with state worth showing (sensors, camera previews, energy data)

The voice-integration plan is part of the product spec, not an afterthought. The integration affects firmware (which capabilities to expose), industrial design (mic-pickup if the product itself has voice), and marketing (which ecosystem badges to use).

If you are integrating voice control into an IoT product — particularly weighing Matter-only vs Matter-plus-skill — we have shipped both patterns.