Our Parallel/Distributed Intelligence Product Philosophy

Our parallel/distributed and intelligent products are designed to bridge the gap between the power of cognitive-based intelligent solutions and the practical aspects of developing real-world applications. Many research-driven intelligent products are based on a one-size-fits-all approach where a narrow set of solutions is productized without any real consideration of how it can be integrated with other equally useful approaches. While it is simply part of the way researchers think to assume that given enough time and resources their research will someday solve all of Computer Science's Grand Challenges, 50 years of intelligent systems research has shown that this is simply not the case. Further, many broader intelligent solutions that have the theoretical strength to solve real-world problems are supported by software implementations that lack any serious software engineering, and therefore, can not support the type and level of processing needed to solve complex problems.

To try to overcome these limitations, we have based our product strategy on three basic principles:

1. Focus our direct intelligent systems research on breadth vs. depth by developing unified theories that allow the dynamic integration of new and existing intelligent solutions.
2. Focus this unification on the inclusion of the work of others at a substantive level by allowing the reuse of as much existing cognitive and computer science research as possible.
3. Take the time to build a powerful and supportable set of tools that allow intelligent solutions to be safely distributed across very large collections of processing elements.

We believe that the NIH (or not invented here) approach to doing research and development might, for a time, make you more famous or wealthy, but it will never provide long-term solutions to real-world problems. Further, relying on as simple as possible software implementations might, at first, seem like a prudent use of limited resources. After all, in internal R&D, no one really expects you to release the software used to test your published theory, but maybe this is why we have been following so many false leads for over 50 years of intelligent system research.