Please enable JS

Our History

Pedram Ariel Rostami, CEO

Sept. 7, 2020, 2:28 p.m.

From the beginning of our journey, we set our goal to leverage the Artificial Intelligent (AI) and Machine Learning (ML) in resource management systems to increases productivity and efficiency at the enterprise level.

To achieve that, we had four significant obstacles to overcome:

  1. Deal with more than a billion heterogeneous data points.
  2. Create classifiers that can work in dynamic conditions while being context-aware.
  3. Build a system that can operate without supervision.
  4. Design an AI architecture that is lightweight, adaptable, and will not require significant cloud computing infrastructure.

Most of these issues are well-known problems in ML and AI, but what made it even harder for us in system and resource management was the diversity and definition overlaps of data based on technologies, disciplines, engineering, and science backgrounds we targeted to serve.

For example, when analyzing a given input, let's take a sentence as part of an executive summary or a job description or an individual's resume or device operation manual; such information needs to be classified for downstream processes to achieve a proper system response. However, the problem is that the very same sentence can be associated with more than just one discipline/category at a time, which results in the execution of multiple cross-functional classifiers, therefore, system responses.

Running a given input on multiple classifiers and their models is not just impractical but counter-intuitive. It requires significant computational power and introduces considerable uncertainty in the results, and that is all besides the time and speed needed to make split seconds decisions. Therefore, there was an absolute need for a solution that can associate a given input to only one model/or classifier without compromising resource, time, speed, and accuracy.

To address this, we adapted context-aware computing models. In other words, we started by constructing an ecosystem consisting of models in the form of rules or decision criteria, based on a prior knowledge by experts to classify a polarity of input text by its contexts. Therefore, the data can only be related to one classifier at a time with the most incredible accuracy possible.

Such an ecosystem is context-aware as it uses the context to provide information or services relevant to the input's primary subject and concept. However, there was another crucial element; we needed a platform that can operate without human supervision and further training after implementation.

As such, we re-designed the system architecture by dividing it into multiple neural domains. Each domain consists of custom-built classifiers with proprietary Neural Network designs and numerical models. In a nutshell, the system acquires non-labeled data and transforms them into labeled multidimensional matrixes for classification to achieve a proper system adaptation without supervision and live training.

After years of continuous research and development, in Jan 2020, we successfully formulated this ecosystem that can understand a polarity of input text on a contextual level and addresses all four above mentioned criteria and requirements. 


We are currently taking this ecosystem even one step further by creating custom wrappers and offline capabilities to be available on any platform, anytime, anywhere. Our proprietary software might be the only readily available package successfully deployed and implemented on off-grid devices. The algorithms' efficiency and innovative nature can power off-the-shelf devices without significant IT or online cloud infrastructure, if any, in remote and isolated locations worldwide.

The latest exhibition of this technology in action is C9Compainon. An AI that is working as our exclusive and personal daily planner, a kind of a superpower that helps ordinary people become extraordinary achievers.

The potential for utilizing this system and its Human-Computer-Interaction (HCI) capabilities is beyond resource management and system administration. Any task that requires and or avails from contextual data classifications can significantly benefit from such a lightweight yet robust ecosystem. Our HCI approach and design can overhaul any app that needs constant human interaction to reduce all intermediate steps into one actionable item.

We are looking forward to expanding our technology and its core engine in other business areas and actively pursuing potential collaborations and synergies. If you like more information about C9 Companion and its capabilities, please visit us at

Lets make the impossible, possible. lets make dreams into our everyday realities.