Introduction

Welcome to TextKit—applied NLP for developers. On this site, you will find documentation about this product and the several APIs we have created to simplify your job of building products that use the power of Machine Learning on text data.

TextKit focuses on the engineering side of things using state-of-the-art language models that have become mature enough to be used on customer-facing products.

Our API tries to resemble the concept of the Swiss Army Knife, where a single tool can offer a variety of functionalities for many use-cases that are not tightly constrained to specialized domains.

Technology

Most of the models we use come from fine-tuning existing transformer-based models provided by HuggingFace.

Although the out-of-the-box performance is usually great on these models, sometimes, retraining is needed and that's where we intervened taking a less scientific but more pragmatic approach of making it work for real-world use-cases, which can be surprisingly different from academia, where most of these models were created.

We also rely on SpaCy and Deepset's FARM framework which are all great tools that make it possible for this API to exist.

Finally, we serve everything using Amazon Web Services.

For more information on what we used to build TextKit, please see our Credits section.

Supported languages

English is readily available at the moment, and Spanish is coming up next.

This does not apply to the Language Detector API, which is capable of detecting any major language provided enough text.

We pledge to give back

We are still early in our journey but we have ideated from the very beginning to give back 5% of our profits to the open-source projects that make it possible for us to be doing this in the first place. We wouldn't be here without them.

Getting started

Our signup process is constantly changing as we adjust for demand. Please see here for its current state.