How Flockx Uses Elastic to Power AI-Driven Social Discovery at Scale

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.