Now here’s a little nightmare for an enthusiastic product manager. Your product requires a lot of data from multiple public websites to function properly, but the time it takes to scrape it all up and clean it into useable form is so onerous that it threatens product viability. That’s exactly the problem Columbia grad school alumni Pratap Ranade and Ryan Rowe faced a few years back when they set up Airpapa.com, a website that told air travelers what movies would be screened in-flight. So they set to work on a solution, and it’s finally been unveiled: Kimono, a smart data scraper.
The Mountain View-based startup sucks up data from the unstructured web using a point-to-click tool that deftly extracts information even from webpages without an API. And eventually, it’s going to allow even non-developers with no understanding of APIs to track the data they need.
Smarter than its predecessors
Tech-savvy product managers will probably be aware that earlier offerings in this field purporting to be “smart web scrapers,” like Needlebase and Dapper for example, bit the dust, unable to grow beyond a niche audience. But Kimono, says Ranade, is different. It’s beginning by building a developer user base and will reach out to non-technical users later once this first phase has been established. The standard approach from the competition has been the exact reverse of this strategy.
Ultimately, the aim is to make data extraction so simple that anyone can do it. Users install a bookmarklet in their browsers once they’ve signed up, and it does the Kimono magic. As soon as it’s clicked, it morphs the site into a special state that lets users point to the items they need to track. Take movie times as an example: once you’ve clicked on movie titles and their scheduled times, Kimono’s learning algorithm sets about building a data model around the selected items.
Heading skywards
Even the more seasoned product managers will concede that Kimono’s versatility is impressive. It can extract data to a .CSV file, to Excel, or to RSS as email alerts. The data can also be dropped as an embed on a webpage with the “Kimonoblocks” feature.
And it’s catching on: in the one month since Kimono’s launch, it’s garnered ten thousand signups and thousands of active users.