Introducing Textile — A machine learning platform for mobile apps.

Today, we are excited to release Textile, a new platform that dramatically simplifies the development of machine learning and artificial intelligence for mobile apps.

Machine learning for mobile apps is challenging. Collecting raw data from mobile apps disconnects the data science workflow from the source of data, often making steps to improve collection and engineer new data features slow and cumbersome. After data has been collected and modeled, normally at the end of a giant data warehousing and ETL pipeline, those models are next to impossible to run in a mobile app because the features they are trained with are not available on the device. These are big problems because the insights gained from this data may make or break the quality and experience of an app.

A Few Useful Things to Know about Machine Learning

Textile delivers high quality feature extraction right on the device. We achieve this by running an advanced feature extraction pipeline on streaming data and delivering the results over both a private REST API and directly on device, in real-time. For data scientists, this is huge. It means that the models they train on their own servers can then be deployed and run real-time in their mobile app, on any device.

Textile saves time by adding value to your data. Most data scientists spend upwards of 80% of their time cleaning data and preparing data for modeling. The cost of this time is huge and it means that you are spending only a small part of your time doing the most interesting and important part of the work. Textile transforms complicated streams (e.g. geospatial data, in-app time-series, and accelerometer) into data science-ready features you can immediately use your favorite tools on. This makes it easy to start understanding what drives users to your app, what gets them coming back, and what’s missing in their experience.

Try it out. If you are a data scientist working on a mobile app (or looking to expand your data science capabilities) and wish you could spend less time cleaning and processing data and more time coming up with interesting questions to ask (and then actually answering them!), come check out Textile or drop us a line and let us know what you are working on.

Stay caught up. If you want to keep up to date, follow us here, on Twitter, or over on Gitlab.

As always, a clap or five means a lot!