By Sara Miller, NoCamels -
Every company has a wealth of data in multiple formats, but accessing it and managing it is a problem that many of them have identified.
A 2022 poll by major US financial firm Capital One found that most organizations reported difficulties in handling their own data. Seventy-six percent of those polled said they experienced difficulties in understanding the data, while almost 80 percent said organizing it posed a serious challenge.
“Without data cataloging, decision-makers struggle to understand what data they have, how the data is used, and who owns the data,” Capital One said.
The Kal Sense platform was created by northern Israeli startup Kaleidoo, a subsidiary of Tel Aviv-based Bynet Communications. It analyzes every single piece of data that a company has, using algorithms to plough through the information to provide relevant insights.
Kaleidoo founder and General Manager Yudi Bar On compares Kal Sense’s AI capabilities to those of the major tech offerings, such as Google’s Gemini platform or OpenAI’s ChatGPT, utilizing the same principles of large language models (LLM) and natural language processing (NLP), only dedicated to making sense of information in the professional world.
But unlike other AI-powered platforms, Bar On tells NoCamels, Kal Sense was developed out of the realization that most data is not what he calls “AI ready.” With that understanding, Kaleidoo’s platform uniquely “pre-processes” the data so that the LLMs and NLPs can be used to analyze it.
The platform is suitable for any company, from healthcare to insurance, and even any governmental department that is looking to analyze any kind of data, Bar On says.
“It’s a way of [increasing] efficiency and getting the maximum quality of the data,” he explains.
“CEOs and companies sometimes make decisions by gut feeling and we want to help them to reach a [facts-driven] decision.”
He says that Kal Sense draws on all kinds of data generated within an organization, not just formal documents, but also information from sources such as text messages, recorded calls, videos and even social media posts.
“If you have a lot of videos and a lot of images in your data storage, you can find exactly what you’re looking for,” Bar On says.
The platform also is capable of optical character recognition (OCR) – the conversion of images of any text, be it typed or handwritten, into a script that a computer can read.
Once the data is amassed, the platform uses it to provide the answers to questions posed by the searcher.
“You want [it] to analyze text, and not only extract the text, but also understand the text, then I can ask questions regarding the text itself,” says Bar On.
“There is a lot of data in any organization,” he says. “And if you ask the chief technology officer and the chief information officer or anyone in the organization, they probably will tell you that they cannot search and they cannot get the value from their data.”
Bar On gives the example of the Israeli Ministry of Environmental Protection, which is one of a number of government agencies he says are already using Kal Sense for data modeling. The ministry uses the platform to analyze information provided by sensors it has placed at locations across the country.
“[It uses Kal Sense] to upload the data from all the sensors in Israel, store it in one place called a data lake and then analyze it regarding noise, pollution or even people throwing garbage from cars,” he says.
The most sophisticated aspect of the platform is the audio conversion, Bar On explains, as it is able to differentiate between multiple voices, perform noise reduction on background sounds and even understand multiple languages, including Hebrew, Arabic and Russian.
All of this is then converted to text for analysis with, he says, 95 percent accuracy.
But not only does the audio convert to text for analysis, it can also provide real time translation between two languages, allowing users to hold a conversation despite not having a common tongue.
Bar On, who spent more than a decade working in R&D for Israeli defense tech company Rafael, says he created the company in 2019 when he realized that the AI capabilities currently being developed were lacking the function of searching all forms of data.
“We see a lot of companies that are focusing on voice, we see a lot of companies that are focusing on video and images, we see a lot of companies that are focusing on text, but we didn’t see the combination,” he says.
“It’s a different technology, it’s unique.”
The platform is solely designed as B2B, and Bar On explains that its level of sophistication and complexity of programming would make it extremely difficult for individual consumers to use.
“It’s not plug and play,” he says. “You need to integrate it [and] you need to work very hard in order to do that integration.”
Last month, the platform was formally launched at a joint event with industry giants Dell and Nvidia, which provide Kal Sense with its servers and processors.
And Bar On believes that almost every professional body would benefit from having the level of access to its hard to process data that Kal Sense offers.
“We call it three-component data that we can communicate with all together,” he says. “This is why we believe it can fit any organization.”
