HORNET.dev retweeted
Since starting hornet.dev, we have been obsessed with building the most efficient retrieval infrastructure for a new kind of user: agents.
Thanks to our partners running some of the largest agent search API workloads, we've made real progress on building the most efficient engine for the new workload.
For our partners serving agent-driven query traffic at scale, it lands as a direct cut in cost to serve compared to legacy engines: more queries per node, and fewer nodes for the same load.
I'm proud of our cracked founders Henning Baldersheim (CTO) and Jon Marius Venstad who have lead this effort. They have spent months on improving rank-safe top-k query processing, low-level posting optimizations, and lots of profiling to find the hotspots from the new workload.
New post on our blog at Hornet.dev: The scaling dimensions of keyword search
Agents are a new kind of search user with longer queries and much higher query volume.
blog.hornet.dev/the-scaling-…
HORNET.dev retweeted
Probably one of the best blog posts in 2026 steve-yegge.medium.com/softw…
HORNET.dev retweeted
Had a blast talking about retrieval with @lateinteraction, @charles_irl and @hwchase17 on the @langchain Webinar!
My slides are now up at
docs.google.com/presentation…
Also go check out the amazing Augmented Language Models from @full_stack_dl fullstackdeeplearning.com/ll…
I'm looking forward to presenting on Information Retrieval in the age of Generative AI.
I'll discuss text representations for retrieval, zero-shot retrieval, and cherry-picked examples of integrating Vespa with @langchain :)
HORNET.dev retweeted
I’ll be in London this week speaking at this event: luma.com/i625g946.
As agents become the main drivers of search, our retrieval stack needs to evolve. I’ll dig into what retrieval needs to look like for agentic systems, and how we should rethink relevance to make them more reliable.
Looking forward to it, and if you’re working on agents or search infra, let's connect!
HORNET.dev retweeted
Legacy search infrastructure is structurally incapable of supporting the loops of agentic reasoning.
I’ve spent many years working on retrieval, ranking, and large-scale search infrastructure.
But the assumptions embedded in that stack simply don't match the reality of modern AI.
Agents don’t issue short keyword queries.
They search in parallel. They iterate. They build context.
They read full documents and trade latency for thoroughness.
If agents are the future of search, we need a new foundation.
So we started building a retrieval engine designed for how agents actually work.
We're optimizing for recall, throughput, and robust context—addressing the specific failure modes of RAG in production.
We're still early, and we're committed to sharing our journey.
@HornetDev
HORNET.dev hosts another of its series of events - this time in London with @LesterSolbakken at the helm together with Charlie Hull, Olena Gorbatiuk and Erik Schwartz. Fully booked, as always. See you there!
I’ll be in London this week speaking at this event: luma.com/i625g946.
As agents become the main drivers of search, our retrieval stack needs to evolve. I’ll dig into what retrieval needs to look like for agentic systems, and how we should rethink relevance to make them more reliable.
Looking forward to it, and if you’re working on agents or search infra, let's connect!
HORNET.dev retweeted
I'm excited to announce our first hornet.dev event in Stockholm on context engineering for agents!
luma.com/g8iiukhc
AI agents need a new retrieval engine.
Our CEO @jobergum shares why - and our motivation for building Hornet, a retrieval engine built for agents
blog.hornet.dev/the-case-for…