Real-time Query Refinement with Control Vectors

Part of The Frontier of AI Search

Hosted by Trey Grainger, Piotr Kobziakowski, and Philippe Bouzaglou

87 students

In this video

What you'll learn

What are “control vectors”?

How to steer query embeddings on the fly toward refined contexts and personalized user interests.

The “vector math” behind control vectors

Applying the well-known “king - man + woman = queen” approach to generate vector embeddings with updated meanings.

How control vectors improve vector search experiences

See how the UX and relevance of vector search are enhanced by integrating control vectors into query interpretation.

See a live e-commerce search experience and code walkthrough

We'll walk though the source code + demo a live implementation of control vectors using Vespa.ai with Vectra embeddings.

Why this topic matters

Semantic search on embeddings can return low precision results due to a lack of explicit filtering (as in lexical search). Can we solve this by letting users directly inject inclusions/exclusions and updated meaning into their query vectors? Piotr, Trey, & Philippe will introduce “control vectors” for vector search, showing the vector math, code, and a demo of a novel e-commerce search experience.

You'll learn from

Trey Grainger

Author, AI-Powered Search

Trey Grainger is lead author of the book AI-Powered Search (Manning 2025) and founder of Searchkernel, a software consultancy building the next generation of AI-powered search. He also serves as a technical advisor at OpenSource Connections.


He previously served as CTO of Presearch, a decentralized web search engine, and as Chief Algorithms Officer and SVP of Engineering at Lucidworks, a search company whose technology powers hundreds of the world’s leading organizations. Trey is also co-author of the book Solr in Action (Manning 2014), as well as over a dozen other publications including books, journals, and research papers. Trey has 18 years of experience in search and data science focused on building self-learning search platforms integrating the most successful AI Search techniques.


Trey teaches AI Search in the course AI-Powered Search: Modern Retrieval for Humans & Agents with Doug Turnbull.

Piotr Kobziakowski

AI Search Engineer specializing in search, analytics & personalization

Piotr Kobziakowski is an AI Search Engineer specializing in search, analytics. He has designed and optimized large‑scale search and recommendation systems for high‑traffic platforms. He is working extensively with modern search stack Vespa and previously Elasticsearch. His experience spans across relevance modeling, ranking, experimentation, data analytics and data‑driven product improvements, merging engineering and business goals.

As an instructor, Piotr focuses on making search, AI, and personalization practical and accessible.

Philippe Bouzaglou

Technical Founder at Vectra, the AI foundation model for e-commerce search

Philippe pioneered the concept of the Social Graph while attending Harvard with Mark Zuckerberg. He has since gone on to achieve significant contributions in the fields of data science and machine learning. Recently, he was VP of Data Science and Machine Learning and responsible for search at Gorillas, the quick-commerce company that pioneered grocery delivery within the hour. He is now the technical founder at Vectra, an AI company creating the foundation model for e-commerce search.

Worked at

Vespa.Ai
company logo
Elastic
Gorillas
Searchkernel
See all products from Trey & Doug

Go deeper with a course

AI-Powered Search: Modern Retrieval for Humans & Agents
Trey Grainger and Doug Turnbull
View syllabus