Elastic{ON} on demand

The more powerful AI gets, the more your data is worth — whether you're an SRE diagnosing what broke and why, a security analyst chasing a threat through millions of events, or a developer building something that needs to understand the world around it.

Agentic AI is the defining 2026 conversation across every enterprise — but BFSI carries the highest stakes, since agents here can approve transactions, flag fraud, and make credit calls with minimal human oversight. This panel confronts the gap between agentic AI ambition and operational reality: Most institutions can't yet observe, audit, or explain what their agents are doing in real time. Featuring new primary research, panelists will debate what production-grade agentic AI actually requires versus what most organizations have deployed today.

At Nykaa's scale, incident investigation once meant manually sifting through millions of logs across catalog, cart, checkout, and payment services. Running Elastic Observability on AWS, Nykaa's SREs now query logs conversationally via Elastic AI Assistant, surfacing root causes in seconds. Join this fireside chat exploring the company's shift from manual triage to AI-assisted incident resolution.

Everyone is using generative AI; almost no one fully trusts it. Stack Overflow faced that gap head-on, adding AI to 60 million community-validated questions and answers without destroying the trust developers rely on. The answer was to lead with search, not generation. This session walks through the retrieval-first architecture behind OverflowAI — Elasticsearch hybrid retrieval grounding Azure OpenAI — and what Elastic on Azure now offers teams building the same pattern.

Agents are only as good as their context. Google Cloud shares how its partnership with Elastic delivers a complete agentic AI stack. Learn how deep integration between Google's agent tooling, Gemini models, and Elastic retrieval turns fragmented, sensitive enterprise data into trusted context without moving data or compromising governance. Learn how customer Apna, India's largest jobs and professional networking platform, uses this architecture to power advanced semantic matching at scale, connecting millions of job seekers with employers while driving substantial growth and platform productivity.

Getting AI agents into production remains difficult. Token costs, relevance challenges, and inconsistent responses keep many from scaling. The root problem is that agents lack business context, and existing ways of providing it are incomplete and brittle. This session shows how Elastic's new context layer solves that by bringing Elasticsearch's relevance and efficiency to retrieving, assembling, and maintaining context for agents across any framework and data source. You'll leave knowing how to manage context, build reliable agents that cut costs and errors, and ship in any client you use.

Network18, one of India's largest media and broadcast organizations, runs on search at massive scale. But as AI moves from experiment to production, a new challenge emerges: How do you govern it? In this session, Rupendra Sharma, chief architect at Network18, shares how the team is extending Elasticsearch beyond traditional search toward AI governance, bringing visibility and control (as well as trust) to AI-driven workflows. Expect a practitioner-focused look at real use cases, lessons from hands-on experimentation, and takeaways you can put to work straight away.

LLMs don't know your business. Every AI app hits that wall, and closing the gap means putting a vector database under your RAG pipelines, assistants, and agents. Elasticsearch now ships pretuned for vector workloads. We'll build vector search on Elasticsearch step by step, indexing with DiskBBQ quantization, wiring managed GPU inference and native models instead of your own embedding pipeline, and tuning hybrid retrieval for sub-100ms latency. Then we'll scale it with _slice across thousands of tenants. Walk away knowing how to ship production vector search without the tuning or rearchitecting.

See firsthand the innovations driving Elasticsearch forward. The new Columnar Mode stores each field once for inverted index, doc values, and source in columnar format. Federated search queries external sources as if they were a local index. With other efficiencies like DiskBBQ vector quantization, these deliver major storage savings, faster ingestion, and fast analytical queries, closing the gap with purpose-built data stores while keeping the search heritage developers rely on. This session covers the why, the wins, and where the platform is headed. If you build on Elasticsearch, this is where it's going.

Discover how Iron Mountain transformed its InSight® DXP platform from a traditional database-based search approach into an AI-ready data platform for modern enterprise search. Discover the key migration decisions, the move toward hybrid retrieval, and how combining keyword-based and semantic search improved relevance, retrieval accuracy, and performance at production scale. Gain practical insights and lessons from building an AI-ready foundation for enterprise data.

So much of the world's information exists in formats other than text. Jina's multimodal embedding models make that content searchable, right alongside text. We'll trace how we got here: from CLIP's modality gap to Jina's Omni models, open-weight models that encode text, images, audio, video, and PDFs into one semantic space. We'll cover how it works, where it shines, and how to use it. Plus, we'll show how our unified inference service makes any model production-ready, as well as consistent APIs across providers and clouds, regional deployment controls, and flexible model choices that cut operational complexity.

SOC teams spend their days working repetitive alert queues, driving analysts to burn out, while attackers move faster than the queue does. This session introduces AlertZero, an initiative to clear the alert backlog so analysts can spend their time investigating. We'll cover the expanded Attack Discovery features and the new Alert Analysis Workflows, which automate triage from classification through case creation while keeping analysts in the decision. You'll also see how agentic workflows enrich alert data and how native-language support makes findings readable for the people who have to act on them.

Qualys, a leading provider of cloud-based security and compliance solutions, will share honest lessons from its journey — cutting through 1,000+ daily alerts to surface only the ones that matter, streamlining SOC operations, and now piloting the next frontier of autonomous security response. Walk away with a practical blueprint for building a tighter, faster, AI-augmented SOC.

Attackers move faster than analysts can open cases. While detect-and-respond matters, stopping threats you can see before execution and hunting for the ones you can't see matter more. We'll show how you how AlertZero tackles both ends. At the endpoint, AI-powered attacks require controls that deny malicious behavior before execution, including prevention architecture and behavioral protections across all environments. On the hunting side, continuous threat hunting ingests your intelligence sources, generates and validates queries against your data.

Elastic Security Labs Threat Command is the dedicated threat research and intelligence team within Elastic. Composed of security researchers, malware analysts, and threat hunters, they focus on tracking advanced cyber threats, reverse-engineering malware, and uncovering emerging attack campaigns. This team has uncovered numerous novel threats, including the Axios npm vulnerability. We'll show you how we turn these findings into detections and protections for Elastic customers.

AI can accelerate investigations, but when applied to incomplete telemetry, untuned detections, and noisy alerts, it amplifies those weaknesses as quickly as it amplifies insight. This session follows a multistage intrusion to show the right order in Elastic: Define data requirements, correlate behavior, test against legitimate activity, and establish ownership before introducing AI. Attack Discovery is then used for context and triage within clear privacy, confidence, and human-approval boundaries. Attendees leave with production-readiness and AI-governance checklists.

Explore the future of security operations, where Elastic's agentic capabilities redefine the endpoint experience. We will demonstrate how autonomous forensic investigation and agentic endpoint operations replace manual, fragmented tool-chaining to accelerate incident response. We'll also show how to achieve AlertZero through pre-triaged, validated intelligence. Finally, we will walk through the stages of a real-world, practitioner-focused threat hunting workflow. Discover how combining native endpoint prevention with agentic reasoning allows analysts to move beyond the alert queue and focus entirely on critical decision-making.

Metrics in Elastic now has native Prometheus and PromQL support, out-of-the-box Kubernetes and AWS investigation experiences (Workflows, alerts, dashboards, skills, MCP app), and automated migration from Datadog and Grafana. We'll cover how the new columnar engine delivers 30x faster queries and up to 2.5x more efficient storage than Prometheus, at 50% of the cost of Datadog. But benchmarks are the floor, not the story. See metrics exploration in Discover, ES|QL subqueries correlating metrics with logs and traces, and agentic investigations that run the moment an alert fires.

Managing over 200 applications across 40+ markets, each with its own regulatory landscape and uptime expectations, is no small feat. In this session, Vidya Vidyasagar, distinguished engineer and global head of cloud and production engineering, wealth & retail banking technology at Standard Chartered Bank, shares how her team is building production engineering as a discipline while driving customer-centric observability across a highly distributed banking technology landscape.

Streams is the unified data pipeline you've always wanted, designed to simplify bringing data into Elastic. Not just logs — metrics, traces, and every other signal, ingested in a schema-agnostic model that replaces the fragmented sprawl of Beats, Logstash, and Agent. We'll introduce the five foundational concepts — Source, Stream, Pipeline, Route, and Destination — that give you one mental model across edge and cloud. See the new visual topology view in Kibana: your entire data flow, end to end, in one map. Then we'll go under the hood into the managed-first processing engine that runs outside Elasticsearch to cut cost and boost performance.

Watch a developer in Cursor generate OpenTelemetry instrumentation live, and then follow the signals to where they land. On a proprietary back end, that open telemetry gets reclassified as billable custom metrics and locked into a vendor collector. This session shows the native path: instrumentation flowing through a managed OTLP endpoint into Elastic's columnar metrics store — no translation layer, no cardinality tax. Once there, one query language spans your metrics, logs, and traces, so correlation doesn't mean switching tools. We'll cover Elastic Agent as an OpenTelemetry Collector and OpAMP pushing SDK changes from Kibana. One open standard, from code to answer.

Find out how Sprinklr operates Elasticsearch at firehose scale with AI-powered SRE workflows using MCP tools. A single enterprise tenant can hold massive social listening volume across a cluster, making cluster health, shard allocation, node performance, and data lifecycle challenging to investigate manually. We'll explore how agentic AI skills use MCP to investigate cluster health, nodes, shards, allocation, and production issues, turning operational data into actionable SRE workflows. We'll also cover the Elasticsearch architecture underneath it, including time-sharded partner-scoped indices, hot/cold allocation, S3 searchable snapshots, and the mechanisms that keep large-scale data searchable. See real-world agent investigations, production lessons, and a practitioner's take on building AI-assisted SRE workflows with Elasticsearch in 9.x.

Welcome to the future of autonomous incident response. We will demonstrate how the Elastic nightshift AI SRE detects and investigates across your entire stack the way your best engineer would. Traversing a causal decision chain, it tests multiple hypotheses in parallel and delivers evidence-based remediation with its reasoning made transparent. We'll also show how complete context elevates the quality of every investigation with cost-efficient storage for full telemetry retention; code-to-telemetry intelligence from your GitHub repos; and precise, relevance-ranked organizational knowledge no other platform can match. Attend this session to get a preview of Elastic nightshift AI SRE in action.