Work on high-speed blackboard systems has lagged behind similar work on rule-based and logic-based systems. Furthermore, there is no common blackboard language to aid in comparative analysis of blackboard architectures. This paper presents a low-level language for blackboard systems. We also present efficient blackboard activation and agenda maintenance mechanisms. We explain why RETE-like pattern-matching networks are not appropriate for blackboard systems. Instead, activation demons provide an efficient mechanism for agenda maintenance. Our results show a significant speedup (52-65%) in agenda maintenance and execution when using the new mechanism in the BBI blackboard architecture. Topic: problem solving efficiency; domain: expert system tools; language; BBI blackboard architecture, Common LISP; status: initial results, compiler still under development; effort; one person-year; impact: a 52-65% reduction in agenda maintenance and execution time.