You are here:
< All Topics
Table of Contents


PAI’s mission is to develop and provide a very intelligent entity that swims in the digital world nativally but yet be very accessible to humans, we chose to build this entity on top of AI based software robot (bot) named PAI-BOT.

The system also supports the most advanced distributed network architecture that will provide the network fully distributed. It enables the system to overcome single point of failure (SPOF) state, increase efficiency by running tasks in parallel mode and automate processes.

PAI uses AI technologies to take distributed computing to the next step – Intelligent bots (PAI-BOT) are being used as an interface to the distributed O/S. 

PAI is a self-learning system based on a collaborative AI knowledge base. PAI leverages Blockchain technology to create the first decentralized brain and nervous system for intelligent computerized entities (PAI-BOTS). 

PAI-BOT uses automation processes to overcome the biggest barriers and challenges of the digital world, like building & running the right distribution platform based on the use-case, collect all the required data regarding the application, load the code modules and AI algorithms from the community code knowledbase (PAI-KB), create all the network elements and servers for the deployment and deploy new bots that will maintain the new application autonomously.

PAI-TECH strongly believes that the world is going to more distributed solutions mostly because of trust, security & privacy issues.


Operating System (O/S) like Microsoft Windows, Linux, Android & iOS and some more are designated to operate the device (computer) that the O/S i installed on. Nevertheless, the system cannot break the device limitation and use other device resources unless using a dedicated synchronization & distribution software. 

If we’ll take two computers running a distributed O/S we will have almost twice better performance, memory, bandwidth etc’, The distributed O/S should not replace the standard O/S but fully control it to operate the device the base way and yet to share processes, data and computing power with the distributed O/S that will be on top of the standard O/S.

The way to control the standard O/S is to install a bot on the system that will be fully managed by PAI-BOSS and will automate and execute processes on the hosting device.


PAI-BOT is as an intelligent interface to the distributed operating system.. 


PAI-BOT is based on 4 major concepts:

1. Intelligent – PAI-BOT can learn with access to a collaborative knowledge base, and share knowledge with other bots. PAI-BOT can learn from, as well as teach another PAI-BOT.

2. Automation – PAI-BOT is a software container that can automate process execution. The process can be written in any language and the PAI-BOT will execute on demand, or automatically when defined.

3. Control – PAI-BOT is always in control and uses a proprietary real-time distributed protocol (PAI-CODE) and an intuitive control panel.

4. PAI-CODE Port – This port enables the communication with the bot – it enables incoming and outgoing messages that contain PAI-CODE commands for execution and bot responses. 



PAI-BOT is based on the following building blocks:


Distributed programming language and protocol that enables one to program and control the bot in real-time. With PAI-CODE bots can be easily upgraded with new algorithms and features.


Module is an extension (application) that the PAI-BOT can “learn” (install) and “forget” (uninstall).

PAI-BOT comes with built-in modules:

PAI-O/S Controller

This module is the physical connector between the PAI-BOT and the device O/S.

PAI-O/S module is responsible for the bot’s security, privacy, scheduling of tasks and other O/S tasks.


This module enables connections to and from the bot. 

Basic connectors that the bot supports are: HTTP, Blockchain, Databases, and Mail (POP/SMTP).

Every module can support multiple drivers; for example, the Database module can support MySQL and Oracle drivers and the Blockchain module will support Ethereum and HyperLedger drivers. 


This module handles all the storage access for the bot. It supports 3 major options:

  • Local Storage – access to the device’s physical storage (the main thread storage)
  • Remote Storage – remote storage with HTTP/SFTP/Samba protocols
  • Decentralized Storage – like IPFS, FileCoin, Sia, Storj & Swarm

In addition, the module supports encryption of the data that is stored using a few encryption methods like Intel SGX or AMD Sev/AMD-V.

Public Artificial Intelligent

This module is responsible for the Artificial Intelligent of the bot (the bot brain).

To support real AI scenarios, the module will include a few AI elements and methods like:

1. Collaborative AI Knowledgebase

PAI-BOT is a self learning entity. To support such a feature, all PAI-BOTS can share public classified AI information between each other and create a rule-based system to ensure the bot works properly. In other words, every PAI-BOT logs its activities in a chain. If an error occurs during an atomic operation or a flow of operations, the system will create a new rule to prevent the other bots from encountering the same error.

The main challenges of such a network is a fast search algorithm and pattern matching.

  • More detailed description about Collaborative AI Knowledge Base and the implementation in PAI-BOT can be found in PAI-BOT Collaborative AI Knowledge Base white paper.

2. AI Context Manager

The Context Manager can handle multiple context sessions for the bot. The context enables the bot to chain a bulk or flow of atomic operations into a single transaction. For example, in order to answer two questions with a direct relationship, in a row, a context is needed:

  a. “would you like to go to eat?”

b. “where?”

These two questions are bonded to each other with a context, which defines that the answer to question “b” is related to the first question.

PAI-BOT uses context manager to run a flow of commands/tasks that need a close relationship and possibly might run different processes and at different times.

More detailed description about AI Context Manager and the implementation in PAI-BOT can be found in PAI-BOT AI Context Manager” white paper.

3. Intelligent Planning

Since PAI-BOT is a self-learning entity it needs the ability to plan ahead. Planning is not just about predefined tasks and scheduling, there might be a need to replan the entire mission or task. For example, we can think of a robot that is being sent on a mission to the second floor, The robot’s task includes going up in the elevator to the second floor, however the elevator is not working (or stuck with the robot inside). The robot now needs to replan the entire task and set a new plan in place with different priorities. 

The ability to replan and set new priorities is called Intelligent Planning, and it’s embedded naturally in most intelligent creatures and defined by a set of rules and logic that can be implemented into a software and PAI-BOT.

* More detailed description about intelligent planning and the implementation in PAI-BOT can be found in PAI-BOT Intelligent Planning white paper.

4. Natural Language Understanding

NLU (Natural Language Understanding) module is responsible directly for the bot accessibility feature. The purpose of it is to translate human language (written or spoken) into a generic context that can be understood by a computer software (the bot). the NLU is used as a parser to the context and flow managers to translate the human language into PAI-CODE.

5. Vision

The Vision module is like the eyes of the bot. It runs the full functionality of computer vision algorithms using classic OpenCV, Deep Neural Networks, Machine Learning etc’.

* More detailed description about PAI Vision and the implementation in PAI-BOT can be found in PAI-BOT Vision white paper.

6. Analysis

The analysis module is used to draw conclusions from big data. Most of the bot activities create a huge amount of data that needs to be analysed in real-time and in post-processing.

A fast and effective analysis method will enable the system real-time performance and better AI features, like finding false negative data in a sequence of events. this will help the bots to improve the ongoing performance using the collaborative knowledge base.