Providing Human Input To Automation: Exceptions overview


Occasionally during the automation process, certain situations may arise that require additional attention or guidance to determine the appropriate course of action. In such instances, Kognitos offers an intuitive interface for users to provide input and guidance to the automation!

With Kognitos, user input plays a pivotal role in enabling the automations to adapt without having to rewrite or adjust the initial automation. This iterative self-improvement process ensures edge-cases and exceptions are not only addressed, but factored in for future automations. By leveraging the provided feedback, Kognitos enhances its problem-solving capabilities, ensuring that future occurrences of the issue can be effectively addressed and the automations get more robust as time goes on.

Feature Highlight: Easy Exception Handling

When your automation requests help from the human, the human can provide feedback and the automation learns and adjusts based off this feedback in future runs. With traditional automation, to handle the same kind of issues in the future, the automation itself needs to be updated, tested, and redeployed. With Kognitos, once the human provides feedback, that's it! The automation can learn from the feeedback and adjust to similar uncertainty in future runs!

How Will My Users Be Notified Their Help Is Needed?

When user input is needed from the automation, an email will be sent with the subject Automation Waiting: I don't know how to proceed. Please answer the question. Within the email, there will be a link that directs a user to the Processes page to address a specific run that is waiting to proceed.

The link will take you to a specific run that will open up directly to the prompt for help to the user from the automation:

Users can also view all processes that need help when viewing all the runs for a process.

For example, when you click on a Process, under the runs tab, any run that needs help will have an orange triangle with the status "waiting for answer" indicating user input is required: