How Flockx Uses Elastic to Power AI-Driven Social Discovery at Scale
Flockx is a social discovery startup that uses AI agents to help people find events, local communities, and like-minded individuals in their area. The company built its core platform on Elasticsearch, using semantic search, RAG, and Elastic Observability to power personalized recommendations and real-time operations. Search response times dropped from hundreds of milliseconds to tens of milliseconds, a 10x improvement, while infrastructure deployment time shrunk from months to days.
Impact
10x
Search response time improvement
Days to hours
Root cause analysis time reduction
Days or weeks instead of months
Infrastructure deployment speed
100,000+
Activation campaign AI requests handled
Challenge
Flockx needed a search and matching backend capable of natural language queries, vector embeddings, and real-time scaling to support AI agents connecting users to local events—capabilities that traditional SQL databases could not provide at the required speed.
Solution
Flockx deployed Elasticsearch as its core platform layer, implementing semantic search with vector embeddings, RAG-powered AI agent responses, and Elastic Observability for pipeline monitoring, all running on Google Cloud with auto-scaling to handle demand spikes.
Tools & Technologies
What Leaders Say
“Elastic is a cornerstone of our platform, powering search across our proprietary data, user information, and AI models.”
“Elastic cuts down on the cognitive load for our teams, from root cause to getting a change out. This used to take up to two days and now it takes less than 24 hours. The ability to create business dashboards directly from Elastic data has been invaluable. It provides real-time insights and eliminates the need for additional monitoring infrastructure.”
“Ease of deployment with Elastic running on Google Cloud has been a game-changer. With a single click, we can deploy our applications to a trusted ecosystem, significantly reducing deployment time.”
“Elastic has significantly accelerated our time-to-market. Previously, setting up our infrastructure and data pipelines could take months and require a diverse range of engineering skills. With Elastic, we’ve been able to rapidly deploy business logic and capabilities within days or weeks, instead of months.”
Sign up to read complete case studies, access detailed metrics, and unlock all use cases.
Full Story
Flockx was founded on a counterintuitive premise: smartphones are making people lonelier, and the cure is to use AI to get people off their phones and into the real world. The company’s platform connects users to local events, clubs, and social gatherings by deploying AI agents called Community AIs that learn user preferences and facilitate introductions. For this to work, the search and matching engine underneath it all had to be fast, accurate, and capable of understanding natural language rather than just keywords.
Early in its development, Flockx evaluated traditional SQL databases as a backend for its search and matching logic. The problem was speed and flexibility. SQL databases could store the data, but couldn’t power the kind of fuzzy, proximity-aware, semantically-rich queries the platform needed to feel intelligent. Every extra hundred milliseconds in a search response is a moment of friction that breaks the user experience—and with social discovery, where spontaneity matters, latency is the enemy.
Flockx chose Elasticsearch as the backbone of its platform and built on top of it extensively. The team implemented semantic search using vector embeddings, allowing users and Community AIs to query in natural language. If a query returns poor results, the agent refines and resubmits it automatically. The company uses Elastic’s RAG framework to pull context from multiple knowledge bases into AI responses. Elastic Observability monitors the embedding pipelines in real time, providing throughput visibility and catching scaling issues before they affect users. Elastic AI Assistant lets non-technical team members query data, debug problems, and build Kibana dashboards without engineering support.
The performance improvements are concrete. Search response times dropped from hundreds of milliseconds to tens of milliseconds—a 10x gain that allows the platform to handle more concurrent requests at lower cost. Root cause analysis, which previously took up to two days, now takes less than 24 hours. When Flockx ran an activation campaign that generated over 100,000 Community AI requests in a single weekend, Elastic’s auto-scaling on Google Cloud absorbed the spike without manual intervention.
Looking ahead, Flockx’s CEO envisions Community AIs operating across wearable devices and smart glasses as costs decrease and adoption grows. The vision is for AI assistants to handle logistics on behalf of users, freeing them from screen dependency entirely. Elastic’s role as the search and observability layer positions the company to scale that vision as the hardware market catches up.