OpenSearchProj

OpenSearch is a community-driven, open source search and analytics suite. Join the conversations at:

[OpenSearchProj @OpenSearchProj]

If faster vector search, reduced storage requirements, and more powerful observability are on your list, OpenSearch 3.9 is worth a look.

Key updates include:
• Neural sparse ANN search with 39% higher throughput, 3.3x faster index builds, and an 8x smaller JVM heap in testing
• Native half_float vectors that can reduce vector memory and storage by roughly 50%
• New 2-bit and 4-bit scalar quantization options for tuning compression vs. recall
• A unified view for alerts, anomalies, and forecasts
• Adaptive per-action concurrency limits for protecting clusters under load
• Index-level search pruning that reduced tail latency by more than 80% and can_match requests by nearly 98% in local benchmarks

Read the blog for the full breakdown of what’s new in OpenSearch 3.9: https://t.co/X6LKqmcHOX

[OpenSearchProj @OpenSearchProj]

We're at AGNTCon + MCPCon North America! Stop by the OpenSearch booth and see how we're bringing observability into the agentic AI stack.

And don't miss Anirudha Jadhav and Shenoy Pratik Gurudatt tomorrow as they break down how a shared MCP architecture connects alerts, traces, logs, and dashboards without losing context.

Oct 23 | 1:45–2:10 PM | LL20A

See you there: https://t.co/FGMann2iO7

[OpenSearchProj @OpenSearchProj]

89% know OpenSearch. But how many know what it can actually do?

According to the 2026 Open Data Infrastructure Report, 62% of OpenSearch users already run AI workloads, yet only 36% consider their deployments mission-critical.

At Open Source Summit Europe, OpenSearch Executive Director Bianca Lewis discussed what’s driving adoption, why awareness of its full capabilities still lags, and what it takes to move from experimentation to production.

Read @Diginomica's latest analysis on OpenSearch’s next phase of growth: https://t.co/5FzJgj5RG2

[OpenSearchProj @OpenSearchProj]

@NVIDIA Nemotron 3.5 Embed 1B is making agentic search more relevant without adding end-to-end latency.

In OpenSearch benchmarks, Nemotron improved retrieval relevance by 65% for neural search and 31% for hybrid search compared to bge-m3, with fewer search requests and similar latency.

Read the full benchmark analysis: https://t.co/eleTczOWCP

[OpenSearchProj @OpenSearchProj]

How do you ingest over 3TB of data daily while reporting zero Severity 1 escalations since 2024?

@Akamai migrated to OpenSearch to reduce licensing costs and gain greater control over its analytics infrastructure.

With proactive monitoring through NeverBlink AI and expert support from OpenSearch Software Foundation member @BigDataBoutique, its seven-person SRE team can identify issues early and resolve potential bottlenecks in minutes rather than hours.

Read the case study to see how they did it: https://t.co/OszwoVIRv2

[OpenSearchProj @OpenSearchProj]

At All Things Open, OpenSearch Ambassador @Horovits took the stage to explore how OpenSearch, Prometheus, and OpenTelemetry are joining forces for better observability, while fellow Ambassador Eric Pugh kept the community conversations going at the booth.

Join the community: https://t.co/yYJqTcq3Uf

[OpenSearchProj @OpenSearchProj]

Your first PR could be your maintainer origin story.

OpenSearch's Contribution Ladder makes it easier to go from first-time contributor to community leader, with a clear path to grow your impact.

Ready to level up? See how to climb the ladder: https://t.co/cDNmmjDnsp

[OpenSearchProj @OpenSearchProj]

At OpenSearchCon NA 2026, we caught up with Emma Gauthier from IBM to discuss how contributors are improving OpenSearch efficiency at scale.

As a member of the OpenSearch Software Foundation and Gold Sponsor of this year's event, IBM continues to contribute to the technology behind OpenSearch.

[OpenSearchProj @OpenSearchProj]

Building an AI agent is one thing. Getting it to remember past interactions and tasks without adding more infrastructure is another.

At Open Source Summit Europe, OpenSearch leaders discussed how storing agent memory alongside retrieval helps developers simplify context management.

But as AI moves from experimentation to production, can organizations scale without adding complexity, cost, and vendor lock-in?

Read ECI Research’s take on OpenSearch and the AI maturity gap: https://t.co/lUjiIIH1G1

[OpenSearchProj @OpenSearchProj]

Ranking #1 doesn't mean a result is relevant, accurate, or safe to act on. And when agents make decisions autonomously, one bad retrieval can trigger a chain of costly mistakes.

Chanpreet Singh and Shatakshi Pandey explore how to build search-confidence guardrails with OpenSearch using confidence scoring, adaptive thresholds, and a critical capability: knowing when not to act.

Because better search results aren't enough if your agent doesn't know when to say, "I don't know."

Read the deep dive: https://t.co/bCo2Mi7zGE

[OpenSearchProj @OpenSearchProj]

Your metrics tell you something broke, but finding out why shouldn't mean jumping between dashboards.

The new OpenSearch data source plugin for Perses brings logs, metrics, and traces into one view.

Available in Perses 0.54.0, it supports PPL queries, automatic time filtering, and server-side credential handling.

Read the blog to see how SAP Cloud Infrastructure is using the plugin to simplify incident investigations: https://t.co/fVe1T2cSQy

By Akshay Iyyadurai Balasundaram, Kelechi Oliver Azorji

[OpenSearchProj @OpenSearchProj]

Her AI agents told her to enter a hackathon. She ended up winning it.

At OpenSearchCon North America 2026, Grand Prize winner Ciki Zeng shared the story behind unclosed, an OpenSearch agent skill built around one question: What if AI had to prove its conclusions before calling an investigation complete?

From building with agents to rethinking how we verify their work, Ciki's experience is a reminder of what independent builders bring to open source.

Read her story: https://t.co/m6s08pWfRG

[OpenSearchProj @OpenSearchProj]

“Search is having a moment.” Now, it’s becoming foundational to AI infrastructure.

From hybrid search and vector retrieval to the context AI agents need, OpenSearch is evolving beyond traditional search to support increasingly complex AI workloads.

451 Research | S&P Global analyst James Curtis returned to OpenSearchCon NA 2026 to discuss what’s changed in the past year and where OpenSearch fits in the future of AI infrastructure.

[OpenSearchProj @OpenSearchProj]

Last week at Open Source Summit Europe, OpenSearch sponsored the event, connected with the community at our booth, and explored agentic search on stage with OpenSearch Ambassadors @Horovits and Aswath N Srinivasan.

Thanks to everyone who stopped by and made it a great week!

[OpenSearchProj @OpenSearchProj]

Alert rules multiplying faster than your services?

See how OpenSearch uses PPL + Unified Alerting to correlate logs, metrics, and traces, centralize routing, and cut false-positive noise without sacrificing coverage.

Watch the live technical demo on demand → https://t.co/TEqgwFKQ25

@TheNewStack @LinuxFoundation

[OpenSearchProj @OpenSearchProj]

Already in Prague for Open Source Summit Europe? 👀

Keep the week going at OpenSearch Prague Meetup #1 tomorrow, October 8.

Search, observability, lightning talks + food and drinks with the community.

📍 Novotel Praha Wenceslas Square
🕕 18:00

Register: https://t.co/5fUQ7jJ1qn

[OpenSearchProj @OpenSearchProj]

Stephanie Walter, Practice Leader – AI Stack at HyperFRAME Research, breaks down how OpenSearch is becoming a retrieval and context layer for enterprise AI.

Because better AI starts with better retrieval.

Read her take → https://t.co/gfzjB74z5I

[OpenSearchProj @OpenSearchProj]

Nine keynotes. One very clear signal: OpenSearch is leveling up for the agentic AI era.

@Aiven_io, Amazon Web Services (@AWS), @CERN, @IBM, @Intel, the OpenSearch Software Foundation, Portal26, @SAP, and @Uber covered retrieval, memory, observability, security, scale, and what it takes to make AI trustworthy in production.

Catch the full recap → https://t.co/rqo2MspF4N

[OpenSearchProj @OpenSearchProj]

Add resource-level access controls without coupling your plugin to Security.

In OpenSearch 3.9, plugin pages can expose sharing controls for detectors, monitors, model groups, workflows, and more through a simple placeholder element.

No import. No plugin dependency. No manifest entry.

Blog: https://t.co/id3f8Shz45

[OpenSearchProj @OpenSearchProj]

@TheBrianGraf came for the search. Stayed for the community.

From his first OpenSearchCon to diving into vector search, hybrid retrieval, and RAG, hear what made Brian want to keep building with OpenSearch and become an Ambassador along the way.

Meet Brian Graf → https://t.co/6tAtJqRiMW

[OpenSearchProj @OpenSearchProj]

We’re grateful to see OpenSearch featured as the Analytics & Data Intelligence Solutions winner in the 2026–2027 SiliconANGLE TechForward Buyer’s Guide!

Thank you to the @thecubetv & @theCUBE team and this year’s judging panel for the recognition.

https://t.co/uB8nCVaj3k.