Reports, videos, audios, emails, PPTs… Amid all your data ‘chaos’, the Flaps system can detect the name, place or information you’re searching for and even in the exact paragraph.
Published by magazine D+I on February the 3th 2021.
How much time did you waste the last time you were searching for that data you needed but you couldn’t remember in which document or presentation you had saved it? This ‘chaos’ often takes hold of many employees at companies operating on knowledge when filing and saving all that prized information and documentation in our digital era.
Using artificial intelligence and natural language processing algorithms, the Catalan startup Flap has developed a system that’s capable of not only accessing non-structured information (text, images, videos and even audios) but also analyzing it. This is known as an insihgt engine.
Thus, irrespective of the file language, this technology can detect names of people and organizations, dates, cities, addresses, names of buildings, legal clauses… “All of this is indexed and, what’s more, all of this information is divided into paragraphs; and the audios and videos are converted to text to extract all the data,” explains Eudald Camprubí, Co-Founder and CEO of Flaps, to D+I.
When users enter Flaps, all they have to do is search for the information they need and the result is “the specific paragraph or minute of video or audio containing the information you’re looking for without having to read it all”. In other words, it locates not only the documents with the information, but also the paragraphs where it’s at.
Plus, this system is also able to do optical character recognition (OCR) to index the text displayed in multimedia content. In short, it cross-checks information even when found in different areas of the company, on Google Drive, on an email server (including attachments), in Dropbox or the local folder system, for example.
Likewise, it also allows companies to connect through their API to synchronize a specific data source used within their organization. To this end, he mentions that they aren’t at this time planning to offer integration with chat services like Slack, WhatsApp or the now-unavailable Hangouts “due to the noise generated when searching for transcendental information”.
To explain how the technology developed works, Camprubí points out that the algorithms “are perfected in the scope of a company’s knowledge at the same pace it analyzes the organization’s information” which essentially means they become “personalized” to each company.
Software programs that help companies generate knowledge “have become rather obsolete because people get lost”; hence, the evolution lies in implementing these types of algorithms developed to “get more performance” out of the information with analysis and indexing.
Specifically, the Flaps system works using vectorization techniques. In other words, everything –the search as well as the results– is converted to numbers so a user’s question, for example, doesn’t have to be based on any of those typical keywords but rather on semantics.
An entrepreneurial evolution
After 12 years working on the development of business document management systems, Camprubí and his partner decided to create a startup, Onna.com, upon noticing there was a “serious problem” with “efficiently managing information in documents of a more or less large size”.
Camprubí explains that standard management systems are basically folders where finding information is a rather “dramatic” process. Through Onna, they were able to reach clients of the size of EA, Facebook and large San Francisco (USA)-based firms in the legal tech sector.
In order to leap over to Europe, they created a new startup, Flaps, aimed at focusing on the Old Continent’s. This took them two years of development work following all their prior acquired knowledge.