AiGENT-TECH develops planning, simulation and operation platforms for urban transportation services. Using deep-reinforcement learning framework and off-line/on-line learning, AiGENT-TECH enables on-demand service providers to run a profitable service. Our planning tools are used today to design and test various ITS strategies, making sure the service will be profitable and with the right QoS (quality of service) and pricing for end customers.

Rolling out a profitable on-demand urban transport service is a complicated task.

Taking into account expected revenues & OPEX while quoting upfront-low prices and providing a suitable quality of service (QoS) makes profitability very hard to achieve.  

Adding 'wild' competition and daily-life uncertainties, turns this mission to be almost impossible for human brain.    

Developing a profitable transport service strategy no longer can be done with traditional engineering tools. The problem complexity and uncertainty impose new methods ... 'optimal decisions making under uncertainty'.  


Our cloud based platform can manage internal-combustion-engine vehicles (ICEVs), battery-electric vehicles (BEVs) as well as future autonomous vehicles (AVs). We put extra effort on BEVs since they improve OPEX and reduce pollution.

Our electrification solution is embedded in each subsystem-module such as BEVs optimal routes, BEVs charging scheduling while keeping QoS ... including electrification considerations in the upfront pricing engine.

In addition, we put extra focus on market competition, since competition is the trigger for so many biz decisions. Our upfront pricing engine is augmented with opponents learning algorithms making sure your service is the optimal for the given market landscape.             


Deep reinforcement learning, probabilistic inference, statistical compact world modeling

AiGENT-TECH develops  long-term expected reward processes and decision making under uncertainty algorithms, known as reasoning and planning. Our algorithms replace the limited rule-based systems that are being used today in so many industries. 

Using deep reinforcement learning, our AiGENTs improve (handling the cognitive tasks better and better) over time based on their own experience. We have our in house discrete/continuous modeling and inference engines. 

Our deep reinforcement learning framework - combination of model based/model free - trains the AiGENTs with a very small data set, which is critical in real life applications. 

Our statistical modeling capabilities is the main reason we can handle "more than Atari games" use cases with our novel deep reinforcement learning framework.

©2017 by AiGENT-Tech.com