India's smart home market is growing fast, but voice control hasn't caught up. Most users rely on Alexa or Google Home-platforms that require setup, speak primarily in English, and have no native understanding of the Havells device ecosystem. For the many households that have already invested in Havells appliances, controlling them by voice means jumping between apps, linking accounts, and learning a new command syntax that doesn't match how they naturally speak.
The brief for Voice Mode was to build an AI powered voice agent that feels native to the Havells One app, works with zero configuration, and speaks the way Indian households actually speak. No jumping between apps. No Alexa. No Google Home. No account linking. Just talk.
Before touching a single frame, we spent 2 weeks auditing how existing voice assistants behave in a smart home context. We looked at Alexa +, Google Home, LG, and a handful of regional players like Atomberg to understand where the experience falls apart for Indian users.
The Multitasking Parent is walking in with a sleeping toddler, hands literally full. She needs to turn on the AC and dim the lights without unlocking her phone or navigating an interface. A direct voice command resolves the whole situation in one breath.
The Elderly User — a grandfather character — is frustrated by small text and complex menus. The breakthrough insight here was language: when he says "bahar ki light on kar do," the system should respond in kind. "Ji, light on kar di hai" isn't just a translation — it's a signal that the technology understands him on his terms
The Allergy Sufferer is returning home after a day in a dusty city and needs clean air immediately. "Ghar ki hawa saaf kar do, bahut dhool mitti hai aaj" — the system infers that what's needed is the air purifier set to fast speed, without requiring the user to name the device or the setting.
Voice Mode introduces an agentic AI assistant within the Havells One app that allows users to interact with their smart home using natural, conversational language.
Instead of manually controlling devices or creating automations, users can simply express what they want to achieve. The system interprets user intent, understands contextual cues, and orchestrates multiple devices seamlessly to deliver the desired outcome.
Static screens are easy to design. What Voice Mode actually needed was a system of motion that communicated three distinct states — idle, active, processing — through changes in the oval's feel without ever being distracting. The principle we kept returning to was "motion with intent." Every transition had to serve a communicative purpose: state change, feedback, or emotional tone.
Voice Mode activation rate (% of eligible sessions where Voice Mode was opened): [X]% Task completion rate (commands resulting in a device state change): [X]% Multi-device command rate (commands affecting 2+ devices in a single session): [X]% Repeat activation rate (users returning to Voice Mode in a second session): [X]%
