Elastic review: Search, observability and security platform built on Elasticsearch, for AI and data teams

What Elastic does

Stores and searches documents, vectors, logs, metrics and traces in one engine for AI retrieval, observability and security work.

Elastic is the company behind Elasticsearch, a distributed search and analytics engine that the vendor presents as a search database for AI. It keeps documents, vectors, logs, metrics, traces and other data shapes in one engine, and it supports full-text, vector, semantic and hybrid search, plus retrieval pipelines that give language models grounded context.

Three solutions sit on top of that engine: Elasticsearch for search and AI applications, Elastic Observability for logs, metrics, traces and APM, and Elastic Security for SIEM, endpoint protection and automation. Kibana is the shared visualization and management interface, and data arrives through Elastic Agent, Beats or Logstash. A long integrations catalog covers cloud services, databases, security tools and workplace apps.

You can run it as Elastic Cloud Serverless, Elastic Cloud Hosted, or self-managed on Linux, macOS, Windows, Docker or Kubernetes. Hosted pricing is resource based and serverless pricing is usage based; both can be paid as you go monthly or prepaid, while self-managed is licensed by number of nodes and used RAM. Pricing not published on the official site: no dollar or euro amounts appear on the pricing page, so you need to contact sales or start the 14-day trial, which needs no credit card. The vendor lists a 99.95% monthly uptime SLA for Platinum and Enterprise cloud tiers, and says the site is meant for professional use only.

• Vector, semantic and hybrid search
• Retrieval pipelines for grounding LLMs
• Observability for logs, metrics, traces and APM
• SIEM and endpoint protection
• Serverless, hosted and self-managed deployment
• AutoOps cluster monitoring and root-cause analysis
• Alerts to Slack, PagerDuty and Microsoft Teams
• Large integrations catalog

Our take

Elastic suits engineering, security and operations teams that want one engine for search, AI retrieval, monitoring and threat detection, and that are comfortable choosing between serverless, hosted and self-managed deployment. The trade-off is that the official pages publish no prices, so budgeting means talking to sales or running the trial. Serverless currently lacks some capabilities that hosted and self-managed offer. The vendor says the product is for professional use only, so it is not aimed at consumers or hobby users who want a simple fixed-price plan.