November 2021 Release 🥳

Our engineering and AI teams at smartlayers have been very busy during the last month working on our second ever version of the platform. In our November version we have implemented a combination of user requested features along with a set of planned product features Mahuna and David have been working on since day 1. In this small post we will go over our November version feature highlights with some short explanation and videos to go along.

Improved Dataset Creation and Annotation

Using our easy to use annotation and dataset creation platform users can annotate their document data easily and accurately.

One-Click Pipeline Deployments

After you have annotated at least 20 documents you can train and deploy our custom document understanding models, called “smart ocr” models with the click of a button. Training time will vary between 15 minutes and 2 hours depending on the quality of the documents and the size of your datasets.

Human in the Loop AI Powered Document Understanding

Once your model is fully deployed you can connect it to any system you need or directly feed documents through your portal. A fully trained and deployed model is called a “pipeline”, you can access all of the information regarding your pipeline from the pipeline UI. AI is not always perfect, but with smartlayers you can have confidence in your models by breaking accuracy down by entity type allowing you to understand which fields are going to be, on average, correctly predicted. Any incorrect result generated by your smart ocr model pipeline can easily be corrected using our custom Human in the Loop process.

Download Reviewed Data

When it comes to using your structured data there are many options our users can choose from. Your dataset can be queried directly as a structured table using SQL like syntax using our API. Users can also directly download your pipeline results into an excel or spreadsheet.

API Usage and API Tokens

Even though our human facing platform is amazing, our backend API is even better. Any action that a user can conduct in our frontend can be triggered using our REST API, allowing for integration with hundreds of programming languages and other SaaS platforms. Securely access your data by creating specific API tokens that can easily be deleted.