Vector Search

Vector embeddings excel at text similarity retrieval but cannot effectively handle structured graph constraints, such as simultaneously satisfying multiple conditions like geographic distance, available ICU beds, on-duty specialists, and allergy contraindications.

Vector representations are compressed semantic summaries that lose details such as exact identifiers, negations and numerical boundaries, and may also rank texts that are semantically similar but opposite in conclusion ahead of others.

Vector search maps queries and documents into dense vectors in a high-dimensional space, measuring semantic closeness via cosine similarity or dot product.

Without a baseline, adding vector search can only be called a "change", not an "improvement"; any newly added technical solution must be able to beat the baseline to count as an improvement.

Vector search compares semantic similarity and is inherently poor at exact matching (such as numbers, dates, and codes); directed queries are a weakness of vector search.

Baruzzo's argument is not against vector search itself, but about the order of doing things: first establish a measurable baseline, and let error types drive technical decisions.

When the evaluation set shows that vocabulary-mismatch errors dominate, hybrid search (keyword plus vector) has sufficient justification, and the benefit can be quantified.

Semantic retrieval (vector search over stored knowledge) and state recovery (loading a checkpointed workflow) solve different problems; vector search can surface relevant context, but should never be the sole mechanism for restoring the exact state of an in-progress task.

Vector search is a technique that encodes unstructured data such as text and images into high-dimensional vectors (embeddings), then uses approximate nearest neighbor (ANN) algorithms to retrieve semantically similar items; common implementations include FAISS, Milvus, Pinecone and pgvector.

Compared with keyword inverted indexes, vector search can capture expressions that are semantically identical but literally different, but it requires an additional embedding service and index maintenance costs.

Vector search alone is powerless for exact matches (such as product numbers, error IDs, version numbers and names); when a user searches for 'error E4021', vector search loses the exact match.

Vector search is the number-one AI feature Microsoft has built for SQL Server 25 and Azure SQL Hyperscale.

In the loan scoring example, the stored procedure first performs a vector search to find similar historical loan records, then constructs a system prompt and user prompt organized into a REST payload.

The core of vector search is semantic similarity, which cannot distinguish hard differences along business dimensions such as year.

Questions with explicit constraints (time, category, permissions) should be resolved first with filtering, while semantically ambiguous questions are handled by vector search—the two complement rather than replace each other.

Pure vector search falls short for information with high requirements for structure and precision (such as names and dates), and needs to be supplemented with metadata filtering

Vector search can retrieve text chunks with high similarity scores, but cannot explain the reasoning logic behind answers, nor structurally connect facts scattered across different documents.

Pure vector solutions excel at open-ended, semantically similar Q&A, but are inferior to graph structures in scenarios requiring precise relational reasoning, cross-source integration, and interpretability.

Vector search works by embedding text into a high-dimensional space and using similarity measures to find the segments closest to the query.

Vector search was not designed from the outset to represent relationships and multi-hop reasoning.