The state-of-the-art of software automation requires explicit programming by a developer. Tools that make this programming easier have gained popularity, but the paradigm is essentially the same: a human must program an automation step-by-step. This intricate, manual programming is time-consuming and error-prone, making automation prohibitively expensive for most of the world's businesses. Can we learn automations using software? Mimica records clicks and keystrokes, and tries to identify repetition in this dataset. Once it learns a map of the process, it can generate code for an automation. We face challenges in .NET, machine learning, and computer vision.
YOU MAY ALSO LIKE:
Automating Automation: Learning by Observation
Tuhin Chakraborty is the CEO of Mimica, a young startup that believes its software can learn to automate work simply by observing it. Before Mimica, Tuhin led engineering teams at companies like Pandora and LinkedIn while building and selling enterprise software in his spare time. He studied machine learning at Stanford University before dropping out to build a startup in London.