Now Reading
How This Startup Is Using Swarm AI To Make Deep Learning Technology Accessible For Everyone

How This Startup Is Using Swarm AI To Make Deep Learning Technology Accessible For Everyone

Ambika Choudhury
W3Schools

Swarm AI is a modern AI technology that is relatively new to organisations. It blends global and local insights to improve and optimise business decisions. Though the concept of swarm intelligence is now new in literature, it is increasingly being used to predict everything from stock market movements to forecasting sales. 

Advances in the Internet of Things technology, machine learning and 5G has made artificial swarm systems faster and more efficient. In today’s world of business that constantly witnesses increasing flux, scale, and complexity, artificial swarm intelligence will help them identify new growth opportunities as well as to anticipate and manage disruption.

With a similar vision, Germany-based startup, Brainalyzed is utilising AI and machine learning to make deep learning technology accessible for everyone. To know more about how the company is democratising AI and developing solutions, we caught up with the founders of Brainalyzed.



Founded in 2017 by Dr Gunter Fischer and Thomas Kopetsch, Brainalyzed is an artificial swarm intelligence platform for finance enterprises and is claimed to be the world’s first artificial swarm intelligence (ASI) platform. The mission of this startup is to democratise as well as create AI solutions that decrease the costs and risks, increase operational efficiency, and help the users make better investment decisions.

How It Enables Swarm AI 

Through Brainalyzed Insight, an industry-agnostic swarm AI platform, it is building custom AI solutions for various use cases and data problems. Fischer mentioned some of the highlights of this platform, as below:

  • Genetic Algorithm automatically designs each AI model 
  • Multiple AI models mimic natural swarm intelligence 
  • Automatically find inputs with predictive value 
  • Dynamic adaption of swarm composition based on prediction accuracy in production
  • Absence of the need for feature engineering
  • High prediction robustness due to model diversification 
  • It can run parallel AI model training in cloud or on-premises
  • It is industry agnostic and does not require prior data or coding experience to use the platform

“The difference in our product lies in how we create AI models,” shared Kopetsch. He further shared that the company is creating models with the aid of automated evolutionary AI design as well as the flexibility of data pre-processing, which gives the user the chance to combine data from distributed sources with the transparency of pricing. It competes with companies such as H2O Driverless AI, DataRobot, Google Cloud AutoML in the domain.

The Tech Behind It

The learning backend at Brainalyzed only uses deep neural networks as the type of machine learning model. According to Fischer, since Brainalyzed Insight is a machine learning application, they have tried to leverage open-source frameworks as much as possible, to benefit from the community development. 

See Also
startups india nasscom

All the major machine learning parts are using Python as a programming language and most importantly is the ML core, which is based on Google’s TensorFlow. The optimisation of inputs and architectures is accomplished with a custom-made genetic algorithm. Besides that, the company is using Pandas for data wrangling and AsyncIO for asynchronous process management.

Furthermore, to spawn and manage cloud nodes within AWS, Azure and IBM, the company is relying on the Hashi Corp stack with Nomad as a cluster manager and Consul as a service manager. On the front-end side, Brainalyzed is building the UI using React for single page application, and Redux for the state management and the design is done with MaterialUI and SASS for the styling.

Future Roadmap

In the next five years, the company is planning to use the training data of the hundreds of thousands of training that are being conducted to improve the process with an AI model itself. As Fischer shares, the startup plans to use the statistical properties of each input and the prediction targets, to suggest initial configuration, which can be further optimised with the custom-built Genetic Algorithm.

What Do You Think?

If you loved this story, do join our Telegram Community.


Also, you can write for us and be one of the 500+ experts who have contributed stories at AIM. Share your nominations here.

Copyright Analytics India Magazine Pvt Ltd

Scroll To Top