WattsUp

Inspiration

Home appliances are smarter, more interconnected, and more energy-efficient than they’ve ever been, but the infrastructure for monitoring and managing its usage has remained stagnant. And for homeowners and renters, there aren’t many resources to self-educate on how to optimally use these appliances to minimize their footprint on the energy grid and to make savings on their electricity bills.

With WattsUp, we’ve created an educational and empowering way to learn about your energy consumption, take proactive actions to minimize your consumption, and foster an environmentally-minded community through friendly competition.

What it does

Once the user logs in via Auth0, WattsUp plugs into their home’s real-time consumption telemetry. The dashboard contains a display of the user’s total energy usage in kWh, actionable tips to help users save on their energy, and a control center for managing and automating their smart appliance activity. All of this information is used to educate the user on their energy usage and guide them to create informed decisions on how to shrink their grid footprint. WattsUp also features a leaderboard full of your friends and neighbours where you can compete for which household has the best utilization of energy saving techniques.

How we built it

We found a dataset containing the records of energy usage, appliance type, time of energy consumption, and associated environmental factors for 500 homes. Using this dataset, we trained the AI to learn the behavior for any given household (user) and provide recommendations on how they could save energy by making small changes to their habits. From this data, we generated real-time insights into the user’s grid impact, translating raw consumption into tangible environmental benefits and direct economic savings. To complement these insights, we developed a device management center that tracks model specifications and warranty status, while introducing a custom "wellness rating" to evaluate appliance health and operational efficiency.

The tip system was carefully integrated with the devices so the user can directly accept the tip and authorize the action. Each tip contained meta data on potential actions as well as the device associated, and once the user accepted a tip, the meta data passed allowed the state of the device to be automatically changed. We also integrated a Vultr database system to hold the different user’s devices, update their status and hold the logs. Older logs were stored in a separate object storage system, holding it in a JSON structure. The final app’s backend was hosted through a Vultr Server.

Challenges we ran into

• We spent a while finding accurate and applicable residential appliance data to use for our project because most datasets were either outdated or lacked the detail we wanted.
• We explored various LLMs including Claude API, but found that they were either far too complex or not optimized for our use case, so we switched to Google’s Gemini API to streamline and focus our building efforts.
• Initially we tracked smart devices with simple on/off toggle controls, but as we integrated more complex devices with variable settings, we had to engineer a more robust UI that could handle diverse controls without overwhelming the user.

Accomplishments that we're proud of

• We successfully used Gemini to take in household data and generate personalized, actionable energy-saving tips tailored to each user’s specific consumption patterns.
• We calculated and displayed values to quantify a home’s total footprint, translating kWh into tangible metrics like cost savings and total carbon emission reductions.
• We developed a scoring algorithm based on energy usage per square foot to ensure that efficiency is measured by performance rather than home size, allowing all size households to complete a level playing field.

What we learned

A lot was learned from all stages of this project. From preliminary background research on the topic we learned the true impact individual appliances have on a household’s cumulative energy consumption, which allowed us to prioritize the most high-impact optimization strategies. During the conception of the product, we refined our skills in prompting and tweaking various LLMs to transform complex, technical energy telemetry into intuitive, natural-language recommendations that any homeowner can act on. Finally, we learned the importance of understanding the end user and creating features that prioritized ease and accessibility while maintaining a seamless and intuitive interface.

What's next for WattsUp

For the future, we would implement enhanced automated scheduling features that allow users to effortlessly shift their energy consumption to off-peak hours.

Built With

• api
• auth0
• csv
• expo.io
• fastapi
• gemini
• json
• postgresql
• python
• react-native
• vultr