AI, VR, Mobile and fare dodging in Prague
Do you know that in Prague, people ride the underground without buying a ticket just to play a game of cat and mouse with inspectors for the fun of it. To stay one step ahead of the inspectors a company built a mobile app using Artificial Intelligence to predict what the inspectors would do next. Not very honest, but quite a clever use of game theory and AI. What I did the inspectors do next? Well hold that thought.
Last week. I attended a technology study tour in Prague. An opportunity to spend two days with a variety of CIOs, Innovation Leads, Architects and others from other sectors and meet the brightest and best of Czech early stage companies. The main themes covered were Virtual Reality, Artificial Intelligence, Mobility and gaining Insights from satellite imagery and unstructured text. There were others, but these are the ones I would class as the main ones.
Space_Know are a Czech company based in Prague and in Silicon Valley specialising in applying machine learning to analyse satellite imagery to aid strategic planning. With the rapid changes in the satellite imagery market, driven by the falling size and cost of low orbit satellites, Space_Know are able to buy access to imagery with quite detailed resolution (30cm per pixel) for more or less every part of the globe, 20-30 times a year. From this they are able to initially train their systems what to look for, and then within minutes, they are able to analyse global imagery over time to look for changes in size or status of certain things such as: levels of oil tanks, sizes of forests and, using shadow analysis, even the size and coverage of buildings. In effect, if it shows up on an image, it can be analysed.
The next theme was the analysis of unstructured data such as social media and text documents including Microsoft Word and emails from multiple sources. Although the core technology has been around for some time, what is interesting is that companies like Geneea and Keboola are starting to make this technology very accessible to a non-technical user, enabling them to simply identify entities such as names, places and even overall themes and then set up rules on how to process or report on that data, such as feeding into CRM systems or flagging dissatisfaction to help desks. Interestingly as with many early stage companies around the global, Geneea and Kebools are doing different things, but working together in a number of projects. We see this open collaboration a lot in the entrepreneurial world.
Next up was mobile security. Although the mobile market had been mature for some time, companies such as TeskaLabs and Avast are still finding niches within which to innovate. TeskaLabs have created a way to secure mobile apps and how they connect to your network using a software defined network between the app and your server. What is especially interesting is the work that TeskaLabs are doing with Cisco to embed their solution into Cisco products to be able to secure the Internet of Things.
Avast, known for their internet security products, have created a way to secure enterprise applications on a mobile device by creating a solution similar to a virtual desktop, whilst allowing the rest of the device to remain open. Their main market is the Health sector, allowing individuals to bring their own device, whilst also giving secure access to enterprise apps without needing to manage the device itself
The VR market has been developing rapidly over the past couple of years with Microsoft developing their Hololens and Facebook buying Oculus Rift, plus many other device manufactures emerging. However, one of the challenges for this market is how difficult it has been to create the virtual environment . This is where Solirax comes in. Although still at the alpha stage, they have created a way to quickly scan environments and import them into their solution, which is offered as a service. This allows people with limited developer skills to quickly create a virtual world and characters.
Finally, several companies including GoodAI, Blindspot Solutions and Professor Michal Pechoucek all presented their work on Artificial Intelligence and Machine Learning. This is a fascinating area and one that I believe we will see advance very quickly in the coming years. Blindspot talked about their work using 'narrow AI' where you use AI tuned to solve specific problems such a logistics scheduling or predicting crime in certain city areas, whereas GoodAI talked about their longer term goal on getting 'General AI' to work in a way that learns as people do, solving a wide range of problems without needing to be trained.
So, what did the Inspectors do next? Well, you won't be surprise to hear that they worked with BlindSpot to build their own mobile app using game theory and machine learning to predict what the fare dodgers would do. And the game continues.