Generating Startups. Randomly.

We use Hidden Markov models and other Natural language proceccing techniques to generate meaningful text for each section of the startup landing page.

Experiments has shown the amount and relevance of data which is used in generating startup text has a significant role in the quality of results. In current experiments we're using real startup submission data to train StartupBot text generation.

Next

Why is this different

Previous attempts have made parodies about the ​randomness​ of startups. Our take is a little deeper, we believe the process of creating a startup can be largely automated.

Next

Most recent generated startups:

Next

About us

Samim Mirhosseini

Co-Creator: Samim is undergraduate researcher at NC State University, currently working on interesting projects like Blink and StartupBot.

Chris Parnin

Co-Creator: Chris is assistant professor at NCSU researching automation in software engineering.