Creating the Intelligence in the Home

At Lexi, we’re transforming the way people interact with their Smart-Home. We believe users want the perfect environment for their homes or workspace - BUT without the burden of having to learn complex setup and commands.

Lexi Learning Machine

Our Smart-Home as-a-Service business model enables unprecedented machine learning, not only in personalization systems, but also in human computation, device management, environmental management, algorithmic lighting composition design and many other areas.

Experimentation and algorithm development is deeply engrained in everything that Lexi does. We’ll describe a few examples in detail as you scroll along.

The Three Phases Of Smart Home Evolution


Smart Home Today: Reactive



Lexi Today: Proactive

Sensors & Algorithms


Lexi Tomorrow: Predictive

Machine Learning

At Lexi we view Smart-Home management as a needs-anticipation problem (and related problems). We start by considering the "state" of each user at each point in time, in each location and for each situation. And then we apply machine learning and predictive analytics models to best provide unprecedented levels of reactive personalization and automation for every user.

ML/AI State Machine Learns/Anticipates User Preferences & Behavior

We keep track of every touch point we have with each user—every device they add, every piece of feedback we get, every composition they choose, when they go to bed, every time they enter a room, how often they change their settings, etc.

With this data, we try to understand users' states and their needs when in those different states. We can then detect changes in state and consider possible triggers. This process by itself can lead to insights that help us keep our users happier.

And once we define and understand states, and detect and understand clients' transitions between them, we can develop state transition matrices and Markov chain models that allow us to study system-level effects.

Pre-Configs out of the Box

Sensor Input Parsing/Ingestion

Location Aware

Contextual Relevance (dinner, sleep, cooking, etc)

Captures/Defines Usage Patterns

Optimal Human Centric (Circadian Rhythm)

Dynamic Environment Compositions

Predictive Analytics for Personalization


Management Automation Of Smart Devices

With users’ state data and usage patterns, we can learn where a user’s environmental preferences fall along a spectrum of smart-home settings.

These latent features can then be used in our mixed-effects models and elsewhere to output automated device settings.

Beyond Classical Collaborative Filtering

Moving our solution even further beyond classical collaborative filtering, and given the wide range of Lexi supported smart devices. we also have a lot of multi-sensor and voice-interaction data to consider: composition choices, movement patterns, and the vast amount of voice NLP feedback and request utterances we receive from users.