NEWFOL

NEWFOL provides a framework for developing and deploying the cognative parts of an AI applications. The framework includes: an interactive tool for preparing and testing custom knowledge bases that understand your proprietary data and programs; and linkable runtime libraries that integrate this data with NEWFOLs reasoning ability. These runtimes can be used in fully embodied autonomous AI agents (robots). Applications that do not need the full functionallity of NEWFOL (e.g., the ability to understand how how to learn everything from their environment through sensors) can use a simpler version of NEWFOL called NEWFOL Lite.

Building an AI Application

Not all AI applications need to have the more subtle subsystems that make up an autonomous AI Agent. For example, it may be a specialized application where complete domain knowledge comes with the application but the programmer wants a user to be able to tell the application how to use a new strategy for solving new class of problems in the domain.

This is a typical scenario where NEWFOL Lite is useful. It allows a programmer to model the domain and use existing tactics to solve problems in the domain. Once it knows about a domain the NEWFOL Lite runtime can use its preexisting knowledge of problem solving to learn about new domain specific problem solving from the user.

Building an Autonomous AI Agent

We think of an autonomous AI agent as a robot, like Data of Star Wars fame. He(?)'s clearly not biological (artificial) can use his encyclopedic knowledge of many things to solve complex problems (intelligence) but what really distinguishes Data from an ordinary computer application is that his understanding and judgements of the world are his own (autonomy).

That property makes his obversations of our world more interesting and useful to us than if he were simply to be something that parrots what he has been told.

This is the critical difference between an 'autonomous AI agent' and an 'smart/AI application. To put this into the current context: IBUKI believes that generative AI (e.g., using LLMs) is only capable of being an AI application and in our opinion can never grow up to be an autonomous AI agent.

IBUKIs goal is to bring you software that will make AI agents commonplace. The question is 'what kind of software is this?'