Israel is dubbed “Silicon Wadi” for its high concentration of technological companies, densely packed in a Middle Eastern land strip. The country is home to many R&D centers of multinational corporations and local startups trying to break into the global market. Just like other startups around the world, Israeli startups have been exploring the bounds of AI applications, sometimes outside of their own borders. Let’s zoom in on some of the Israeli AI startups whose innovations are making it internationally.
1. Navigating the Waters with Pools of Big Data
Windward, an Israeli startup housed in Tel Aviv and Paris, owes its beginnings to naval experiences at the Israeli army. Some of the company’s founders leveraged the knowledge they’ve acquired during their national service, and are now doing AI by tracking and analyzing shipping behaviors at sea.
The company uses satellite information to identify the location of ships at sea—treating the information as a pool of unorganized big data. It cleans the satellite data to gain actionable insights into what ships are carrying, where they’re heading to, and what the most optimal shipping routes are for them. It also contrasts local data against global shipping patterns for comparison.
The more information Windward can gather about a ship’s described purpose, actual stock, destination, and route selection, the better it can filter down to greater levels of predictive accuracy. It can also uncover data correlations that may be relevant for commercial actors and other stakeholders worldwide.
2. Out with the CFO, in with a Chief Financial AI
For those who are tired of, or just don’t have time to do financial paperwork, Zeitgold has created an app to take care of individual bookkeeping. The startup has R&D roots in Israel, benefits from international investors, and is currently live in Germany, where its app is mainly being used by small businesses.
Zeitgold’s app is responsible for both accounting and financial decision-making. By keeping a retrospective repertoire of financial activities, the app can provide users with suggestions that make financial sense. The ability to gather new insights from past information is what makes the app an AI-driven technology, rather than just being a bookkeeping application.
Clients who want to maximize on the AI potential of Zeitgold’s app can be sure that they own a large database of well-cleaned financial activity data. In this way, when the app’s algorithms are powered to inform clients of possible financial losses or gains, the algorithms have more to work with.
3. Connecting the Dots of Public Transportation Systems
Public transportation systems are complicated beasts, costing governments lots of money. They also have the added challenge of needing to remain affordable for passengers. Optibus is simplifying public transportation’s complexity by optimizing public transportation with AI. Once an Israeli based startup, Optibus now also has an office in New York City, with partners in the United States and Europe.
Optibus gains control over public transportation movements, by integrating the static schedules of public transportation systems with dynamic variables affecting these schedules. Dynamic variables could include traffic, but also the weather or number of available drivers. Optibus instantaneously changes the existing schedule, adapting it to circumstances. In doing so, time can be saved and costs reduced.
The Software-as-a-Service platform that is developed by Optibus makes use of machine learning algorithms, translating the variables that can affect existing schedules into new scenarios. Based on the likelihood of a scenario, an existing schedule can be modified into a more optimal version.
4. From Mammal Vision to Autonomous Vehicles
It started with research into the neuroscientific mechanisms of the rat’s visual cortex at an Israeli university. Today, Cortica has offices in Tel Aviv and New York City, and specializes in true artificial intelligence—mimicking human intelligence. Cortica replicates human vision. Based on research, the visual cortex of mammal and human brains works predictively. Through the eyes, we process visual stimuli, drawing patterns between objects, as our brain quickly calculates what’s coming next.
With insights into the visual cortex, Cortica has developed the kind of AI that’s built on unsupervised learning. For an unsupervised AI, feed the AI with visual stimuli from the outside, and let the AI draw patterns to foresee more information about its environment. This is unlike supervised learning for AI, which works in a top-down kind of way. What limits supervised learning is that the way AI copes with incoming data depends on a predefined strategy. But given unsupervised learning, an AI application should learn to handle new data on the fly, gradually building its own repository of knowledge.
Cortica’s unsupervised AI technology is meeting the advent of self-driving cars and watchful security cameras. Because unsupervised AI doesn’t require human guidance from A to Z, a car could learn to drive autonomously, adapting to the ins and outs of different regions around the world. Equally, a security camera can learn to detect suspicious behavior that is specific to a certain location and context.
5. Teaching Search Engines to Think like Humans
Twiggle, an Israeli startup with a second office in New York City, is looking to change the logic of e-commerce product searches. The startup sees value in semantics, as it wants customers to search for products using conventional language, rather than being limited to commonly recognized search terms.
Regular search engines come with a fixed set of search terms that are matched with a list of products noted in a catalog. Search results depend on specific wordings and on the frequency of particular words having been matched with certain products. No matter how much customers enrich their search queries, if there is no predefined match between a word and a product, then the advanced query is unhelpful.
Using a combination of AI techniques, like natural language processing, enables Twiggle’s technology to conduct a semantic analysis of product searches and results. When users write or verbalize a sentence, possible keywords are identified and related to products and their descriptions. Search results are adaptive to how search queries are written, allowing for greater flexibility. Twiggle has built its AI technology hand-in-hand with an API to support integrations with existing search engines across the web.
But There’s More
Israel’s AI landscape is broader than the startups and verticals that were covered in this article—there’s more. When it comes to AI in healthcare, you can read about Medial EarlySign’s identification of genetic markers and Zebra Medical Vision’s analysis of medical scans. For AI in quality control, you can read about Dragontail System’s double-down on handmade pizzas