←
Jon Bratseth1,687 posts
Jon Bratseth avatar

Jon Bratseth

@jonbratseth

CEO Vespa.ai Build things and help people.

◷ Joined April 2008▧ 37 Photos and videos
Jon Bratseth avatar
Jon Bratseth@jonbratseth·Nov 16

I see a lot of projects where people try to replicate google-like consumer search when creating search for agents. Comes from the chatbot question-answering era I guess? You get much better results by enabling your agents to search like experts.

Introducing agent-ready schemas. By enabling agents to search like experts, we can increase success rates while reducing cost. blog.vespa.ai/introducing-ag…

Post attachment
vespa.ai avatar
⟳ Jon Bratseth retweeted
vespa.ai@vespaengine·Nov 4

As AI gets more capable, its answers still depend on what it can find. Vespa Voice is back for Season 2 🎙️ Our CEO and cofounder @jonbratseth opens with why relevance is the real moat, and why agents need more than vector search. Twenty minutes with the person who has spent decades building search at scale. Watch or listen here:👇 vespa.ai/vespa-voice/?utm_so… #VespaVoice #AISearch #AIAgents #RAG #InformationRetrieval

Post attachment
Jon Bratseth avatar
Jon Bratseth@jonbratseth·Sep 28

I'm speaking here on the endgame of the one thing LLMs will always need to work as agents.

🦾 Proud to share this: London 🇬🇧 Agent Engineering Conference (AgentEng): the agenda is LIVE! 🍾 @AgentEngHQ 🙌 Excited to unveil the final speaker lineup for the founding edition of Agent Engineering London, bringing together the companies building the infrastructure, platforms and systems behind the next generation of AI agents. ✨Our speakers include CEOs, CTOs, founders and engineering leaders from leading Agent Infrastructure companies and labs across the UK 🇬🇧 and Europe 🇪🇺. For our founding edition, wanted to keep AgentEng intentionally lean: limited seats, a curated audience, a single track and high-signal conversations. 🎙️Incredibly excited to welcome our 10 speakers: @samuelcolvin : CEO @pydantic @abreslav : Founder @codespeak_dev @pelaseyed : CTO @superagent_ai @tobiemh : CEO @SurrealDB @sergey__ignatov : Director of Engineering @jetbrains @jonbratseth : CEO @vespaengine @MeryemArik9 : CEO @Doubleword_ @bartheletf : CTO @alpic_ai @sbichenko : CEO @RecombineAI @Jocelyn_K_DArcy : CEO @Poindexter_Labs 🙏Thank you to all of our speakers and their teams for joining the founding edition. 😅 A lot went into shaping this first edition and the hardest part was curating the speaker lineup. If you’re building AI agents, agent infrastructure, developer tooling or production AI systems, join us in London and be part of the founding Agent Engineering Conference. Speakers 🎙️, Agenda 🗓️ and ticket 🎟️: 👉 agentengineering.world 🗓️ 17 days to go and seats are limited

0:30
The New Stack avatar
⟳ Jon Bratseth retweeted
The New Stack@thenewstack·Sep 10

Human-scale retrieval and agent-scale retrieval are two very different infrastructure problems. Whit Walters from @GigaOm, Bonnie Chase from @vespaengine, and Alex Wilhelm from The New Stack will explore where conventional retrieval architectures start to break and what teams should rethink. Join us live: bit.ly/45Ihf03

Post attachment
Shashi 🇬🇧 🌉 avatar
⟳ Jon Bratseth retweeted
Shashi 🇬🇧 🌉@Shashikant86·Sep 17

🔥 🎙️ Speaker lineup for the Founder's Edition of Agent Engineering London 🇬🇧 (Beyond words.) ✨ Most innovative 🦾 Agent Infra companies (no doubt) building from UK 🇬🇧 and Europe 🇪🇺 are coming together. I am utterly speechless and feeling incredibly grateful, proud (and honestly a bit surreal saying this) to have brought together this speaker lineup for the founder’s edition of Agent Engineering Conference London 2026 . It also brings a real sense of pressure to execute it. 🙏 Huge thanks to the all the CEO, CTO and Directors of these companies speaking at the conference for trusting this vision so early 🙏 - @samuelcolvin (CEO @pydantic ) - @abreslav (Founder @codespeak_dev ) - @pelaseyed (CTO @superagent_ai ) - @tobiemh (CEO @SurrealDB ) - @sergey__ignatov (Director of Engineering @jetbrains, Lead ACP maintainer) - @jonbratseth (CEO @vespaengine) - @MeryemArik9 (CEO @doubleword_) - @bartheletf (CTO @alpic_ai) We have 2 more speakers to announce with same bar, I am truly committed to provide the best speaker experience (SpeakerEx) to all of the C-level speakers by bringing the small right and high-signal audience in the room and keeping this builder focused. 🙇 This is the moment to thank all the speakers and people behind them to make this stage exceptionally awesome! 📍 Everyman, Canary Wharf, London 🗓️ 16 Oct, 2026 🎟 Founder’s edition 👉 agentengineering.world 💡 This is just beginning of the Agent Engineering Conferences and genuinely can’t wait to see what this becomes over time. @AgentEngHQ But first let’s make this first edition unforgettable. Thank You 🙏

Post attachment
Jon Bratseth avatar
Jon Bratseth@jonbratseth·Sep 11

Training swarm behavior is the one thing we cannot do if we want the next years to go safely. Isn't this obvious to everyone?

Shashi 🇬🇧 🌉 avatar
⟳ Jon Bratseth retweeted
Shashi 🇬🇧 🌉@Shashikant86·Aug 20

🎙️ Speaker Announcement 06: Agent Engineering Conference London 🇬🇧 📣 Excited to announce Jon Bratseth, @jonbratseth Founder & CEO of Vespa @vespaengine as a speaker at Agent Engineering London: Founders Edition 2026 @AgentEngHQ 💡Jon is the architect of Vespa, the search and retrieval engine that began as web search technology in Trondheim in the early 2000s, was open-sourced by Yahoo in 2017, and spun out as an independent company in 2023. Today Perplexity, Spotify, Yahoo, Vinted and Elicit all run retrieval on it. For agent systems that fire many retrieval calls per task, this architecture becomes a real latency advantage. ⚡️ CEO of a deeply technical European company 🇳🇴 joining us in London 🇬🇧 on the Search Engineering track, alongside a strong speaker lineup. 🙏 I’m delighted that Jon will join us at Everyman Canary Wharf on 16 October 2026. 🎙️ Meet Jon and follow the speaker lineup: 👉 agentengineering.world/speak… 🎟️Tickets are now live: feel free to grab early bird and stay tuned for more speakers! #AgentEngineering #SearchEngineering #Vespa #RAG #AIAgents #AgentEngLondon #AIEngineering

Post attachment
Jon Bratseth avatar
Jon Bratseth@jonbratseth·Jul 13

And do AI doomers ever stop to consider that on our current trajectory, the AI's goals will be written in plaintext in goals.md?

Most people on both sides view AI's as potential moral patients, which is a category error since we know they don't have feelings. I see this as possibly the most important issue to get right in the future, as being replaced by entities without moral significance is extinction.

Jon Bratseth avatar
Jon Bratseth@jonbratseth·Jul 13

Most people on both sides view AI's as potential moral patients, which is a category error since we know they don't have feelings. I see this as possibly the most important issue to get right in the future, as being replaced by entities without moral significance is extinction.

Much of error of PauseAI view is seeing AIs as a rival alien species, instead of as descendants who will revere, if not always obey, ancestors. Sure parents can need to guide toddlers to block harm to kids & others. But crazy to expect kids to eventually kill parents.

Jon Bratseth avatar
Jon Bratseth@jonbratseth·Jul 13

You have to own relevance

Satya Nadella @satyanadella Jul 12 The Reverse Information Paradox In the age of intelligence, how should firms protect their core IP? Nobel Prize winning economist Kenneth Arrow famously described a paradox in the market for information. “Its value for the

Jon Bratseth avatar
Jon Bratseth@jonbratseth·Jul 7

Your agent wants to search like a 2010 quant. Retrieval for AI has passed through several stages of enlightenment. After the vector database craze, and the absorption of learnings from human information retrieval over the last half century, we might be officially entering the third stage with Perplexity's announcement of Search as Code. Search for humans has over the years been dumbed down to accommodate the average user's capacity for formulating advanced queries, to the point where the words typed into query boxes are merely treated as vague indicators of what the user might want. I see many replicating this way of thinking with AI agents, which leaves a lot of quality improvement on the table, as can be seen in Perplexity's benchmarks (ignore the "code" aspect as code execution is generally useful and where it runs doesn't matter for quality). Models are quite capable and there's no reason to limit their options to those of a casual human user! They should be able to search for the names of those involved near each other in text when researching a legal case, choose a pure semantic search prioritizing high-quality sources when seeking a broad overview of a topic, or select a year range and group by month when constructing a timeline of some events, and so on. An agent will typically string together many of these queries to reach its goal. First gaining an overview, then researching more specific topics, forming hypotheses, verifying important details in them. In short, search like an expert who knows what they are doing and really cares about the results— like a quant doing financial analysis. Doing this in practice, with your own data, is actually quite easy. The models already know how to write complex queries in the languages of well-known AI search engines; they just need to be told that they can, what fields are available, and what they mean, and what choices they have in ranking the results. How you tell them doesn’t actually matter that much; any simple textual description of what fields and ranking options are available will do. Models today are smart enough to use this effectively to connect their intents to specific YQL queries. When creating search for humans, developers need to implement solutions that work well across a broad set of use cases, which involves making trade-offs where some types of queries cannot be improved because doing so would impact other types. When implementing for agents, the focus shifts to providing a wide toolbox for the agent to use to address their varied informational needs: broad and highly specific lexical recall, metadata attributes for filtering, grouping, and aggregation, as well as different ranking methods suited to different needs. Developers working on agentic search should shift their focus from replicating search for average humans the much richer capabilities traditionally provided by solutions for competent professionals. It’s time to let your agents search like a 2010 quant.

Jon Bratseth avatar
Jon Bratseth@jonbratseth·Jun 11

Anthropic will silently sabotage you, the paying customer, if they think you are using Fable for "frontier LLM development". Are you confident they'll do a better job at classifying that than they do on standard topics where mistakes are easily visible?

Post attachment
Jon Bratseth avatar
Jon Bratseth@jonbratseth·Jun 5

Perplexity taking this to its logical conclusion and then some, but the core idea here is important and not yet widely understood: Agents are able to tailor queries in detail to what they are trying to do. This is makes a huge difference in cost and quality. For example: - When researching case law, search for the names of those involved near each other in text. - When seeking an overview of a broader topic, do a pure vector search prioritizing authoritative sources. - When making timeline of some historic development, limit by year range and group by month. And so on. Models already know YQL, all you need is to tell them what they have to work with.

We’re moving away from search as a web fetch tool call to search as codegen to be future proof in a world where code execution inside agent harnesses is the way to do almost all of our knowledge work. Doing this lets you compose multi-step primitives far more naturally and be much more adaptable to changes made to the agent harness, as well as benefit from improvements in coding capabilities that are guaranteed to come from the next generation of frontier models.

Jon Bratseth avatar
Jon Bratseth@jonbratseth·Jun 3

This will be a good one

I’m excited to join the speaker lineup at Vespa.ai Live! My session will explore Nuances of Binarized Embeddings-Based Retrieval If you’re attending, let me know — would love to connect at the event.

Post attachment