Extracting information from the content of documents, videos or audio files can be an extremely time consuming and daunting task. This is why we have created the Ask Engine; a unified integrated interface where users can acquire information efficiently and effectively, from multiple sources, by simply asking a question.
For documents, our engine allows a user to simply ask a question about the content and the Ask Engine extracts the relevant information from the document to answer the question. This is made possible by training Ask Engine on millions of documents and pages all over the web to the extent that it can understand the closeness between “sofa” and “couch” or between “beer” and “wine”. So if a book talks about “Bob who drinks wine on his sofa every Friday” and the user asks “who was drinking beer on the couch?” from a 1000-page book, the user will still obtain the most relevant answer “Bob” despite asking an inaccurate question.
For videos and audio, Ask Engine first transcribes the spoken words into text and then treats the generated text as a document, seamlessly letting users ask questions about the media and instantly navigate to the part of the video that has their requested information.
We enable users and businesses with large amounts of opaque and unstructured data, like video surveillance footage, to generate contextual insights. The corpus generated by Ask Engine can be easily queried through natural language or simply used to trigger routines or other actions. For instance, through nFlux, Alexa could alert a mother in the kitchen when her child is climbing out of her crib or a port security manager to instantly query surveillance footage to answer questions like how many containers were offloaded in the past 12 hours.
Understanding lingual and visual signals can open up applications in a variety of fields; entertainment, education, security, and healthcare to name a few. Here we focus on examples from two of the verticals.
Safety & Security
- Setting alarm for an unknown license plate (or person) in front of the house
- “Jaime picked up the package yesterday at 3:14 PM”
- “Luciana is climbing up her crib upstairs”
- Notification that garage door has been left open unattended
- “5 workers in the danger zone are not wearing a helmet”
- “Machine number 6.62 is leaking”
- “Storage unit has shipped 89 containers in the past week”
- “Male with a gun walking in area 25R at LAX”
- “3 men fighting on the intersection of 40th and 13th street”
- Extracting information from large amount CCTV footage
- Ms. Smith left the stove on unattended for the past 20 minutes
- Garry fell and has been unresponsive for the past 7 minutes
- “It’s been 4 hours since you took your medications”
- Patient X needs to retake medications in 19 minutes
- Patient X slept for 7 hours and 20 minutes last night
- Patient X has been exercising for the past 11 minutes