Definition
The performance gap or reasoning failure that occurs when an AI Agent or RAG system lacks sufficient historical context, indexed documents, or few-shot examples to ground its initial operations. In RAG, this often refers to the 'empty index' phase or the trade-off between deployment speed and the time required to generate high-quality embeddings for large datasets.
In AI, this refers to data and context scarcity, not the latency of spinning up serverless containers.
"An intern arriving at a new job with an empty desk and no handover notes, forced to guess until they gain institutional knowledge."
- Few-shot Prompting(Mitigation Strategy)
- Vector Indexing(Prerequisite)
- Context Window(Resource Constraint)
- Semantic Cache(Component)
Conceptual Overview
The performance gap or reasoning failure that occurs when an AI Agent or RAG system lacks sufficient historical context, indexed documents, or few-shot examples to ground its initial operations. In RAG, this often refers to the 'empty index' phase or the trade-off between deployment speed and the time required to generate high-quality embeddings for large datasets.
Disambiguation
In AI, this refers to data and context scarcity, not the latency of spinning up serverless containers.
Visual Analog
An intern arriving at a new job with an empty desk and no handover notes, forced to guess until they gain institutional knowledge.