Form automation
with AI

What we do

Forms still cause a lot of manual effort. Be it for applications, information letters, practice visits, meter readings or returns processes. Thanks to its Deep-OCR, form automation recognizes all text information, including handwriting. A wide variety of form types can be quickly created and automated. The software delivers real-time results, reduces manual effort, requires no setup costs and produces immediate savings. In addition, the entire process is easy and quick to integrate.

Plug & Play

Thanks to Rest API, our AI models are quickly integrated into existing systems.


Our AI models have been trained with hundreds of thousands of documents and extract all important information with high precision.


Thanks to high-performance GPUs, our AI processes documents in real time.


Our OCR supports various languages of the Latin alphabet.

Active learning

Automated processes are constantly improving the recognition rates of our technology.

Robust reading

Our Deep OCR reads even low-quality documents and can be specially adapted to difficult use cases.

Data protection

We are 100 % GDPR- and Schrems II-compliant. We process the data on our servers and dispense with service providers.

AI Boost

Our AI models directly enable your software to automate documents.








+49 151 687899


M. Mustermann

  "processing_id": "bfb819fb-b193-42fa-ad19-8ad33aa8831b",
  "available_results": [
  "pdf": "/processing/results/bfb819fb-b193-42fa-ad19-8ad33aa8831b/pdf",
  "ocr": {
    "pages": [
        "width": 2380,
        "height": 3368,
        "bboxes": [
            "id": 1,
            "x1": 82,
            "y1": 333,
            "x2": 102,
            "y2": 351,
            "text": "a",
            "text_entropy": 0.58203125
            "id": 2,
            "x1": 132,
            "y1": 1508,
            "x2": 551,
            "y2": 1558,
            "text": "Staatsangeh├Ârigkeit:",
            "text_entropy": 0.141357421875
            "id": 3,
            "x1": 152,
            "y1": 108,
            "x2": 5551,
            "y2": 1558,
            "text": "Aktives Mitglied in der Sparte",
            "text_entropy": 0.161421875
Show more +

How we do it

Our technology has been trained on thousands of font samples and can accurately read and automate even wrinkled documents thanks to the newest AI technologies such as the use of Natural Language Processing and Computer Vision. We also draw on knowledge from other document types, minimizing the effort required to create the AI model for each form.
using Newtonsoft.Json.Linq;

using (HttpClient httpClient = new HttpClient())
    string baseURL = "";
    string workflow = "invoice_extraction";
    string apiKey = "<API_KEY>";    // TODO: Insert or load your API-key secret here
    string filePath = "<FILE_PATH>";    // TODO: Insert or load your file path here

    using (HttpRequestMessage request = new HttpRequestMessage(new HttpMethod("POST"), baseURL + "/processing/" + workflow + "?include=extractions"))
        request.Headers.TryAddWithoutValidation("accept", "application/json");
        request.Headers.TryAddWithoutValidation("Authorization", "ApiKey " + apiKey);

        MultipartFormDataContent multipartContent = new MultipartFormDataContent();
        ByteArrayContent file1 = new ByteArrayContent(File.ReadAllBytes(filePath));
        multipartContent.Add(file1, "file", Path.GetFileName(filePath));
        multipartContent.Add(new StringContent("{\"language\": \"de\"}"), "parameters");
        request.Content = multipartContent;

        HttpResponseMessage response = await httpClient.SendAsync(request);

        string responseString = await response.Content.ReadAsStringAsync();
        JObject responseJson = JObject.Parse(responseString);
        if (responseJson["extractions"]["vendor"] != null
            && responseJson["extractions"]["vendor"]["name"] != null)
            JToken vendorName = responseJson["extractions"]["vendor"]["name"];
            Console.WriteLine("Vendor name: " + vendorName["value"].ToString());
            Console.WriteLine("Vendor name confidence: " + vendorName["confidence"].ToString());
            Console.WriteLine("Vendor name could not be extracted.");
Show more +
Show more +
Show more +
Show more +
Show more +

Rest API

Can be integrated into existing processes in just a few minutes, thanks to support for common programming languages such as C#, PHP, Python, Java or NodeJS. Our documentation easily guides you through the integration processes. We support both synchronous and asynchronous data transfer. 
In the first step, documents can be uploaded via the API. Our software supports a whole range of file formats. These are currently the image formats JPEG, PNG, BMMP, GIF, TIFF and the document format PDF. 

Einfach Integration dank Make-Plugin

Easy integration thanks to Make-Plugin

Intuitive use of our technology and easy integration into existing processes such as in the DocuWare context thanks to our Make plugin.

Data extraction

We provide the key information as a structured JSON format via the API. For each data point, it also specifies the exact position on the document, the value and the confidence level. In addition to the JSON file, we provide a PDF/A file with the text level.
Ready for automation?
With you get state-of-the-art AI technologies for intelligent document processing.
Request a meeting and get a free product demo!