SRI's vision of Spoken Language System (SLS) middleware is to
take advantage of available INFORMATION and COMPUTING resources
even though they may be distributed and heterogeneous
This graphic shows a need, and pieces of an existing network of compute servers and data resources
called into service on the fly
to meet this particular need
Systems need to be flexible, dyanically reconfigurable and robust
users may have only a small PDA locally
recomposability (what is done locally, what is done locally) is essential
users needs may be complex and will certainly vary