All document extractions via our web portal or through APIs are counted as 1 API call.
On free plan, your first 100 calls are on us for you to try out FormX.ai and feel free to contact us if you need help.
We only save the master document uploaded to our web portal, or the documents you uploaded for training a custom models.
Any images uploaded via API or web portal test are in-memory only and diposed once extraction is done. We will not use your document for optimising our AI models or any other purposes unless you request us to do so.
All API calls are encrypted with HTTPS.
Yes! We works with Enterprise to customise the plan that suits their needs. Please contact us!
Yes! We provide tailor-made SLA commitment for Enterprise Clients, please contact us!
For Business Plan, the minimal throughput is 180 RPM (Requests per Min) for API call without Product Line Item, or 30 RPM for API call with Product Line Item, without a SLA commitment.
FormX.ai is hosted on GCP and Azure cloud.
Accuracy rate depends on a lot of factors, and it also depends on how accuracy are defined.
As a reference, most of our previous clients would have at least 70% of accuracy by fields.
If we limit to only images with good quality (no blur, no lighting problems), from our client’s feedback, general fields such as date, time & amount can have accuracies over 90%.
You can contact us with around 100 receipts as sample, and we might share an accuracy report with you.
Usually it is 4 - 8 seconds, excluding the image uploading time. (It is not a SLA commitment)
Image uploading time will depend on the image file size and network speed, which we have no control on. We suggest you to resize the images to a smaller one to reduce the uploading time.
If you need a better latency or a SLA commitment, please contact us for an Enterprise plan.
We suggest to assist end-users to take photos of long receipts without blur, with a pano mode alike camera.
We can provide SDKs for clients who needs help on it.
Here are some common reasons for poor accuracy:
- Watermark or chop overlapped with the information we want to extract
- Blurred or glared images
Please send us a message and we will help you to get all your questions answered
