Engineering Blog

Deep dives into vector search algorithms, distributed systems, and the future of Aspected.

Large Language Models

Retrieval Augmented Gen

Search Algorithm

Vector Database

A flowchart illustrating how retrieval failures lead to LLM hallucinations. The process starts with a User Query, followed by a Retrieval Layer failing due to incomplete context or misaligned ranking. This leads to the LLM Generation stage where extrapolation is forced, resulting in a “Hallucination” where the root cause is actually Retrieval Failure.

Feb 19, 2026•Aran Montero Salvadó & 3 others

It Wasn’t a Hallucination. It Was a Retrieval Failure.

An analysis of the root causes of hallucinations in RAG systems, and a relevance-first approach that reduces them by design.