Butler uses recent advances in NLP in combination with OCR and Computer vision to understand documents fully, rather than just locating text in specific locations. This helps capture the information you want accurately, whether it's seen a similar document before or not. You no longer need a team of data entry specialists, and you no longer need to maintain flaky templates for every changing variants. We typically achieve 95%+ accuracy for most types of documents
Yes! We pride ourselves on how easy our API is to use. Simply upload documents and get the extracted results you care about in structured JSON. Integration can take as little as 15 minutes to get up and running.
We created our pricing model to make it as easy as possible for you to try the product out, and then scale up as your business grows. We have a free tier, along paid tiers that have a monthly fee along with usage as described in our pricing page.
Our main focus is on helping you make a custom model tailored for your exact use case and requirements, and doing it quickly! Most of our users have unique extraction requirements and prefer the flexibility of those custom models. We also have a number of predefined models for common use cases in limited access today. Some examples include: Invoices, Receipts and ID Cards (Drivers Licenses and Passports). If you’d like to learn more about those, simply schedule some time with us and we’d be happy to get you set up with them as well.
With our team of data entry experts, Butler Ops, there’s no need for labeling 100s of documents to train your custom model. Simply define what you want extracted, label a handful of examples, and Butler Ops will handle all the labeling and training from there. We’ll also learn from each additional document uploaded to the system automatically, ensuring your custom model will become more accurate over time.
Yes, absolutely. Handwriting is one of the most notoriously difficult parts of document extraction. Butler’s unique OCR ensemble approach, combined with Butler Ops, enables human-like extraction of even the messiest handwriting.