The Noughty Fox

Take better care of your pool Poolex was a quick project to help pool owners take better care of their pools. It would know all the chemicals you need to add, when to clean the filter, when to change the water, and many more. Industries: Chatbot Services: Mobile Team: Android, iOS Country: France Year: 2024 Duration: 1 month Pool Assistance: Having a pool in the backyard is a dream for many people, but taking care of it can be a nightmare. Challenges CHATGPT: Making the integration smooth and real-time RELEVANCE: We wanted to make the responses as relevant as possible Solutions SOCKETS: We used sockets to make the chat update real-time and quickly ASSISTANT: We used a customly trained assistant to   Read More

AI and AR-powered Digital Twins Vera transforms buildings into intelligent digital twins with embedded AI and real-time data, creating environments where people and devices operate seamlessly. Vera enables context-aware applications for efficient operations and personalized, engaging experiences in office buildings, retail stores, healthcare facilities and much more. Industries: Indoor Navigation Building Management Services: iOS, Android Team: 1 Android, 1 iOS Country: Israel Year: 2022 Duration: 2.5 years Challenge PERFORMANCE: Performance was a big challenge. Running ARKit in 60fps, rendering and communicating with a webview, and processing each frame on-device is very resource-heavy. RENDER SYNC: Syncing the ARKit (camera) frames with the webview frames was a challenge as we wanted the content to move in sync with the world. ASYNC STATE MANAGEMENT: Managing the state of the ARKit, webview, GRPC connection, many background services, and the UI was a major challenge. SDK: Creating an SDK that would allow other developers to create their own experiences on top of Vera was a challenge. Solutions METAL: We ended up using Metal for rendering, and after off-loading many operations on background threads, we managed to get from 13fps to 60fps. FRAME SYNC: We implemented a custom frame sync mechanism that would not render the ARKit frame until webview wouldn't finalize it's pipeline, and it worked like a charm. THE MALL: We ended up implementing a custom asynchronous architecture called 'The Mall' on top of Combine and Coroutines that would fully manage all the state of the app. THE MALL: We ended up building Native iOS and Android SDKs, and a React Native wrapper on top, along with web SDKs.   Read More

Social Training Journal Wildy is a social training journal that helps you track your workouts, share them with friends, and get inspired by others. It's a fun and engaging way to stay motivated and reach your fitness goals. Industries: Fitness Services: Mobile Team: Design Country: Moldova Year: 2021 Duration: 1 month Challenges TEMPLATES: Creating generic enough workout templates that would fit all types of exercises was a challenge SHARING: Sharing workout templates with people, letting them edit it and keeping the history was difficult.   Solutions GENERIC DATA MODEL: We ended up creating such a generic data model that even a metric type was generic.  VERSION GRAPH: We ended up implementing a version graph for the templates, so that people could see the history and reuse specific versions. Read More

Conquer your city Walk out the door, put the phone in your pocket and re-discover your city. Compete with friends and people from your city for the biggest territory! Project scope The project involved building a real-time application that would allow users to walk around the city and conquer territories. The app had to be engaging, easy to use, and visually appealing. It should work smoothly on all devices, and displaying that many elements on a map is very challenging. Challenges PERFORMANCE: Displaying that many elements on a map while keeping the smooth scroll is very challenging. WALKING DETECTION: Detecting accurately whether the user is currently walking or not proved to be a big challenge. CLUSTERING: Clustering users' walking paths on into something tangible on a map was a challenge we struggled to solve for a long time. Solutions MAPBOX: We ended up using Mapbox and their advanced rendering algorithms to make sure the app runs smoothly on all devices. SENSOR PIPELINE: We built a pipeline involving the accelerometer, gyroscope and GPS data to accurately detect the user’s activity. H3: We ended up using Uber's H3 library to cluster the paths, and it solved many other issues in the process. Read More