Vespa AI and Surpassing the Limits of Vector Search

Vespa AI and Surpassing the Limits of Vector Search

Author: Software Engineering Daily May 12, 2026 Duration: 38:35

Vector search has risen to become a foundational tool in modern search and retrieval systems, including the RAG pipelines that power many AI applications. However, the demands on retrieval systems are growing more sophisticated, which is revealing the limits of relying on a single vector similarity score.

Vespa is a popular open source search and data serving engine. Central to Vespa’s architecture is tensor-based retrieval, which is an approach that represents data as tensors rather than simple vectors. Tensor-based retrieval enables richer mathematical operations and more flexible ranking functions that can surmount the limitations of a single vector similarity score.

Radu Gheorghe is a software engineer at Vespa with a background spanning nearly 12 years of consulting and training on Elasticsearch and Solr. In this episode, Radu joins Sean Falconer to discuss why vector similarity alone falls short in production, how tensor-based retrieval generalizes to support richer ranking functions, the trade-offs in chunking and multi-stage re-ranking architectures, and where AI search is headed next.

Full Disclosure: This episode is sponsored by Vespa.

Sean’s been an academic, startup founder, and Googler. He has published works covering a wide range of topics from AI to quantum computing. Currently, Sean is an AI Entrepreneur in Residence at Confluent where he works on AI strategy and thought leadership. You can connect with Sean on LinkedIn.

Please click here to see the transcript of this episode.

Sponsorship inquiries: sponsor@softwareengineeringdaily.com

The post Vespa AI and Surpassing the Limits of Vector Search appeared first on Software Engineering Daily.


Dive deep into the conversations shaping the future of artificial intelligence with the Machine Learning Archives-Software Engineering Daily. This curated collection pulls from the broader Software Engineering Daily library, focusing entirely on the intricate world of ML engineering. Each episode features a detailed, technical interview with engineers, researchers, and architects who are building the systems behind today's most advanced AI. You'll hear them break down complex topics like model deployment, data pipeline challenges, and the practical trade-offs involved in taking research from a notebook into production. The discussions are grounded in real-world implementation, moving beyond theoretical concepts to explore the tools, failures, and successes that define the field. For developers and technical leaders looking to understand the nuts and bolts of applied machine learning, this podcast offers a valuable archive of knowledge. It's a direct line to the practitioners who are solving hard problems every day, providing insights you can apply to your own work. Tune in to gain a clearer perspective on how machine learning is integrated into modern software, one detailed conversation at a time.
Author: Language: en-us Episodes: 50

Software Engineering Daily
Podcast Episodes
SED News: Restricted Models, IDE Wars, and the DeepMind Mafia [not-audio_url] [/not-audio_url]

Duration: 51:58
SED News is a monthly podcast from Software Engineering Daily where hosts Gregor Vand and Sean Falconer break down the biggest stories shaping software engineering, Silicon Valley, and the broader tech industry. In this…
Grafana’s Approach to AI-Native Observability [not-audio_url] [/not-audio_url]

Duration: 48:29
Advanced software systems have long been more complex than any single engineer can fully understand. Observability is the established solution to this problem, but with AI agents now generating code, deploying changes, a…
Building Software That People Love [not-audio_url] [/not-audio_url]

Duration: 46:26
Building great software always involves technical problem solving, but the best software goes beyond function. It feels fluid, coherent, and genuinely fun to use. This quality lives at the intersection of engineering and…
Mina the Hollower [not-audio_url] [/not-audio_url]

Duration: 43:48
Yacht Club Games is the studio behind the acclaimed Shovel Knight franchise. Their latest release is Mina the Hollower, which is a top-down action RPG inspired by classic Zelda and Castlevania titles. After many years in…
Foundation Models for Structured Data [not-audio_url] [/not-audio_url]

Duration: 44:00
Predictive modeling is a core element in modern systems, and powers capabilities such as fraud detection, loan approvals, and recommendation systems. These systems typically operate on structured, relational data stored…
Biome and the Future of JavaScript Tooling [not-audio_url] [/not-audio_url]

Duration: 1:02:56
Modern web development requires an ever-growing collection of tools including formatters, linters, bundlers, and plugins. Each tool typically has its own configuration, dependencies, and performance cost. As applications…
Preparing for Q-Day [not-audio_url] [/not-audio_url]

Duration: 46:18
Most of the cryptography securing the internet today rests on mathematical problems that classical computers cannot solve in any reasonable timeframe. That assumption is now being tested. Recent advances in quantum compu…
Developing Multiplayer Games in Godot [not-audio_url] [/not-audio_url]

Duration: 47:21
Multiplayer games are among the hardest software systems to build, requiring developers to synchronize state across unreliable networks while maintaining fairness, performance, and a responsive player experience. Latency…
Web Native Game Development [not-audio_url] [/not-audio_url]

Duration: 54:08
The web has quietly become one of the most capable platforms for game development. Advances in WebAssembly, WebGL, and WebGPU have given developers tools that rival native desktop performance, while game engines like Uni…