Our HOOK Model
How AI "fills the gap"
Backend Integration Diagram
Central A/C isn't a privilege everyone has. In the US, products like the Nest Thermostat are smart and efficient, but also expensive and only work with central A/C. Take Thailand as an example: It's a small, developing country that's HOT year-round. Most people don't have central A/C or smart thermostats, and because of that an average home spends about 65%(!!) of its total electricity bill on window A/Cs alone...
Hive fills the gap by using AI to make your dumb A/C smart!
What it does
Hive uses AI with reinforcement learning to maximize efficiency and comfort from your non-central air conditioner. It also tracks your energy savings and energy usage to provide you with a more complete picture of your behaviors. You start by installing the required hardware: Raspberry Pi, Smart Meter, IR Converter (completely DIY, btw) and then telling Hive to run your A/C in Eco Mode.
Additionally, Hive comes with a community aspect. For example, the skill lets you compare your energy saving to people in your area and it ranks your performance against them. We wanted to make energy saving stupid easy and fun. So we're not going to bombard you with data that you don't understand (kilowatt hour, what?). Instead, we provide a conversation-like energy report, using relatable terms to help people actually understand their impact on the environment.
How we built it
Front end: We used the Python Alexa Skills Kit SDK to develop the Lambda for Alexa. We used Dialog Management and Entity Resolution to handle the more conversational pieces.
Backend: Cloud server running on Microsoft Azure that provides a REST API to receive IoT data through IoT Hub.
Challenges we ran into
- Creating a (mostly) voice-only product with external triggers
- API integration, especially when attempting to query large datasets from Lambda
Accomplishments that we're proud of
- Essentially created cheap Nest for people who needed it most
- If 75% of homes in Thailand with window A/C units were to adopt Hive: 32 Billion Thai Bath == 1 Billion USD in energy savings per year. Plus 4,635 MMTCDE less to worry about!!!
What we learned
- Learned a lot about voice UX, and the HOOK model
What's next for Hive
- Introducing rewards through incentives (let users gain points to claim rewards from company partnerships) and sharing progress over social media
- Using the raspberry pi box as a “push notification”. Green LED is lit when Eco mode is on. Otherwise, it’s red.
- Integration with Alexa Scenes & Alexa Routines
- Letting users share their energy saving achievements on social media