How to start

    How ThinkOwl’s AI works

    Say sayonara to traditional software. Power your service desk with AI.

    Independent of language and structure: ThinkOwl analyses the content of all incoming messages. 

    Start your trial


    Using partially patented algorithms, the system recognizes similarities. ThinkOwl’s AI keeps gaining a higher degree of efficiency based on agents’ implicit feedback on extracting and processing relevant content data.

    ThinkOwl learns

    AI and human-machine collaboration simplify your service operations enormously.


    The principle of reciprocal intelligence: ThinkOwl continually learns from your service experts’ reactions. And your experts are constantly supported by ThinkOwl. As a result, everyone grows a little smarter every day.

    Finally a software that thinks for itself

    Easier to manage – preciser results


    Customers feel appreciated
    because your service finally runs smoothly.


    Agents feel like rockstars because they are empowered to resolve cases faster and simpler.


    With 360-degree intel on productivity and customers management is confident about the future.

    How to get started

    For ThinkOwl to precisely identify your routine customer service processes and intelligently extract relevant data for further processing, you need an adequate pool of sample cases – and a little patience.

    Think big – start small

    • Choose a small expert team of, let’s say, three employees
    • Focus on one communication channel (e.g., email)
    • Together, pick the two or three topic categories that come up the most (routine questions, such as updating customer data, delivery date or the like)

    Match up processes and content

    • Daily define the series of tasks ensuing from your inquiries: What content data is captured from the emails in order to process the customer inquiries? Which responses are you giving repeatedly?
    • Use ThinkOwl Designer to create appropriate input masks for essential content data (so-called “case forms”) – learn more here.

    Training by examples

    The most important step is the initial training of each category through message examples.

    • Send customer cases to your ThinkOwl inbox. Ensure your expert team assigns the cases to the right categories by “click” and transfers the required content data to the input masks.
    • ThinkOwl begins to recognize content after only a few examples. After 80 to 100 exemplary customer emails in the same category, the recognition gets really good. Important: Be sure NOT to choose emails with a mix of topics.
    • Of course, you can manually set conditions in ThinkOwl, too. Learn more here. However, this is not required for your AI test.

    Test with additional emails

    • Now you are ready to send more emails to ThinkOwl. You will notice ThinkOwl automatically identifies customer emails – as long as they fall into one of the trained categories – and autonomically begins to assign the required content data to the appropriate input fields.
    • Note: Simple training shows notable effect. Nevertheless, wont reach the recognition rates that make ThinkOwl so exceptional until the system is actually running the AI.


    We’re happy to support you during your free trial.

    Ask about AI