AI Redefined (AIR) today announced Cogment 1.0, an open source human-AI orchestration and collaboration platform that provides world-class AI development capabilities to everyone, through a robust, future-proof, cost-effective platform that provides marked improvements in multi-agent and hybridization support, training, and computing distribution. Cogment is an open source technology initiative designed to democratize robust, sophisticated AI training capabilities, and is free to use under Apache 2.0 license.
This press release features multimedia. View the full release here: https://www.businesswire.com/news/home/20210622005427/en/
Cogment is AI Redefined’s open source technology initiative designed to democratize robust, sophisticated AI training capabilities, free to use under Apache 2.0 license. (Photo: Business Wire)
Developers and AI researchers are invited to join the GitHub Cogment community today by visiting: www.cogment.ai
Cogment’s flexibility and unique human-AI interchangeability feature enables new, real-time applications that were not possible before. It can readily combine multiple approaches and agents to quickly, reliably, and inexpensively provide solutions to the most specific problems, from prototype to training at scale, all the way to real-time operational deployment.
Human-AI augments; it doesn’t replace people or the unique structure, processes, and operations of an organization. Cogment’s modular AI systems can be grafted onto existing infrastructure and workflows, as gradually or quickly as desired.
With Cogment’s AI Apprentice approach, AI agents must demonstrate their competencies before they are entrusted with the ability to act: through progressive human-in-the-loop (HITL) training, they gain increasing responsibilities -and autonomy – based on their proven abilities.
The human-centric core of Cogment is not only the safest and most ethical approach to onboarding AI; it also ensures a richer understanding, and lowers the risk of bias, flaws, and catastrophic failures, like the ones often seen in the AI-as-Automation systems commonly offered.
Cogment is highly adaptable and scalable for distributed infrastructure. It supports all bleeding-edge AI technologies – and the ability to hybridize them – in a microservice architecture that’s endlessly customizable and future-proof, seamlessly evolving alongside ongoing technological innovations in AI. Cogment’s lean nature and strong performance make AI technology more accessible, while delivering results fast, with lower development costs, less downtime, and less staff retraining.
Thales, a global leader in digital and “deep tech” innovations such as connectivity, big data, artificial intelligence, cybersecurity and quantum computing, leveraged AIR’s experience and expertise – embodied today in Cogment – to develop a solution for the company that is centered on training. “With AIR, we found a partner who would quickly understand our vision for man-machine and machine-machine teaming while allowing us to significantly accelerate our roadmap and unlock new challenges,” Antoine Fagette, Head of AI Research & Technology, Thales.
“Human-AI collaboration technology is too vitally important to our economy and infrastructure – our society – to be confined to proprietary models and corporate interests,” said AI Redefined CEO Craig Vachon. “It fundamentally belongs in the hands of the open source community where it can grow and flourish with ethical guidance.”
Cogment is lightweight and compatible with all commonly-used tools, such as OpenAI Gym, TensorFlow, Keras, PyTorch.
- The Orchestrator, the heart of a Cogment app that orchestrates all the components
- Python and Javascript SDKs which are used to build Cogment app services and clients
- A command like tool that facilitates the creation of Cogment apps
Cogment’s main features include:
- Tech stack agnosticity
- Multi-method implementation
- Multi-actor interactions
- Multi-reward multiple reinforcement learning
- Humans in the loop
- Multi-experience learning
- Human/AI actor hot swapping
- Custom and hybrid agent architectures
- Distributed computing & training
- Iterative, live development
Vachon continued, “AIR will continue working tirelessly to contribute our expertise to the Cogment platform, taking inspiration from the countless developers committed to lending their talent and innovation for the greater good.”
About AI Redefined
AI Redefined’s (AIR) vision is to perfect the synergy between humans and AI to face the challenges of securing a brighter, more ethical future together; its mission is to enable humans and AI to trust and continuously learn from each other via a dynamic human-AI orchestration platform that supports humans at scale, from prototype to operational deployment. Its flagship technology platform, Cogment, is a high-performance, open source framework that allows multi-agent, multi-reward, continuous live training of AIs and humans together in the same environment. For more information, visit http://airedefined.com. Follow us on Twitter @AI_Redefined and LinkedIn ai-r.
View source version on businesswire.com: https://www.businesswire.com/news/home/20210622005427/en/