What is Optimate AI?


Optimate AI is a software platform

for self-driving industrial operations

Optimate AI applies unified modern deep reinforcement learning technology and simulations to train an AI agent to control optimally and autonomously any industrial process. The core technology is production-ready and safety aware, which are crucial factors for any industrial operations.

Key technology distinctions




By using modern Deep Reinforcement Learning technologies, Optimate AI can steer any industrial control or planning process


Solves many problems with unified approach

Don't waste time to design solutions for every new, even slightly different problem. Define your goals and Optimate AI will do the rest


Seamless integration.

Deploy anywhere

Run your process autonomously by deploying Optimate Agent on your IoT platform, cloud or directly on edge devices


Automatically adapts to changing circumstances

Optimate AI constantly analyzes your processes. It will automatically find and apply the new best policy in case of any changes



human expertise

You can explicitly define what Optimate AI will do in certain situations. It useful in rare cases like process start or shutdown


Awareness of safety, steadiness, and consistency

Optimate AI ensures that every produced agent won't violate any constraints and able to steer your process safely

How does it work?

Step 1. Connect Optimate AI to Simulation Engine


Optimate AI finds optimal control policy or plan by safe learning on synthetic data in simulations.

Furthermore, to reflect the real-world processes more accurately and behave robustly in the wild, Optimate AI adds actual process historical data into the mix.

So it needs access to both these sources.

Simulation is an approximate imitation of a real-world process we can interact with and run it with different conditions.


Think about a computer racing car game – it is a simulation of a real-world racing tournament.


In Industry 4.0, simulation that accurately reflects the real-world process is called Digital Twin.

Step 2. Define the goals, constraints, and actions of your process


Tell Optimate AI what you want to optimize for in your process, and what situations you want to avoid. Then Define the additional constraints you don't want to be violated (like safety or regulatory). Finally, define what Optimate AI can take control of in your process.

All these are done by mapping data and simulation parameters to goals, constraints, and actions – artifacts of Optimate AI.

Step 3. Let Optimate AI find the best policy


Start the learning and sit back while Optimate AI learns an optimal policy for your process over millions of different simulations without exposing any of your physical assets at risk.

Under the hood, Optimate AI learns to control your process trough trial and error, very similar to how children learn new skills.


To capture the nuances of the real-world (like noise in sensors) and adapt control or planning policy to production setting, Optimate AI also uses real data.

In Artificial Intelligence, we often call the control or planning policy An Agent.

Typically, simulation time required to train an agent is equivalent to 10-30 years of real process operations. Thus, every Optimate AI agent's experience is equivalent to a seasoned operations expert, but it is acquired in days and weeks, not years.

Step 4. Deploy and run your trained agent


Almost there! Your agent is ready to be deployed to your IoT platform, cloud infrastructure, or the edge device. At runtime, the agent receives real-time data, determines which actions should be applied, and sends it to the actuators. You can choose metrics of the agent to monitor on the dashboard as well as alerts you wish to receive.


Moreover, Optimate AI continually monitors changes in your process and updates the agent accordingly.

And that's it! 

Your process now runs in autonomous mode

Want to see it in action?

Any questions?

We will be delighted to conduct you a demo

Thanks for submitting!

We'll contact you soon