Cloud provider Servercore launched PGVector — PostgreSQL for AI applications
The Tech AI
Get the key points in seconds - a short summary from The Tech AI.
Servercore launched PGVector — a managed PostgreSQL-based database with the pgvector extension pre-installed. The service allows storing vector data alongside relational data for AI scenarios without a separate database, supporting HNSW and IVFFlat indexes. The solution includes automatic scaling, backup, and monitoring via Prometheus, as well as the ability to migrate existing databases.
Servercore, an international IT infrastructure provider, announced the launch of PGVector — a new database management system in its cloud database lineup. The service is a managed PostgreSQL with the pgvector extension pre-installed. Companies can store vector data alongside relational data and use it for semantic search, RAG, and other AI scenarios without a separate vector database.
AI applications increasingly work with vectors — numerical representations of text, images, and documents that reflect their meaning. They allow searching not for exact word matches, but for data most similar to the query in meaning. This is achieved using pgvector, an open-source PostgreSQL extension for storing vectors and performing nearest neighbor search. It supports exact search as well as HNSW and IVFFlat indexes, which accelerate search on large data volumes.
In PGVector, the extension is available in every database created within the cluster; no installation or configuration is required. The familiar PostgreSQL data model is preserved: vector search can be combined with SQL filters, and documents, their metadata, and vectors can be stored in a single database.
The service is suitable for AI assistants and chatbots for corporate documentation, catalog and knowledge base search. AI agent developers can use PGVector as a memory and knowledge store, including in conjunction with orchestrators like n8n, as well as as a backend for MCP servers. For startups and teams just beginning to work with AI, PGVector allows avoiding architectural complexity and sticking to a familiar stack.
PGVector is a managed service: there is no need to independently install and maintain PostgreSQL and pgvector. Cluster configuration can be tailored to workload requirements in terms of vCPU count, RAM, and disk space. The cluster can be scaled as the project grows, and fault tolerance is ensured through replicas. For cluster management, backup, PgBouncer connection pooling, and monitoring with metric export to Prometheus are available. An existing PostgreSQL database can be migrated to PGVector using logical replication or a dump.
Ismoiljon Iminokhunov, Project Manager at Servercore
We see that companies in Central Asia are increasingly adopting AI agents and semantic search in knowledge bases. Many teams already work with PostgreSQL, so we made vector search available on a familiar technology. There is no need to deploy a separate vector database, and businesses can launch AI applications without lengthy setup.
PGVector is already available in the Servercore control panel.
About the company
Servercore is an international IT infrastructure provider with a local presence in Uzbekistan, Kazakhstan, and Kenya.
The company supports businesses at all stages of infrastructure management — from designing IT systems to operation and scaling. A team of experienced engineers provides infrastructure support, allowing clients to focus on core business tasks and product development.
– 6+ years of experience in Central Asian markets
– 250+ custom projects per year
– 1000+ active clients from various industries
The company is a resident of the international Astana Hub technology park — the largest innovation hub in Central Asia — as well as IT Park, established in Uzbekistan to stimulate interest in information technology and support promising startup projects.
The Tech on Telegram Follow The Tech on Telegram
Keep up with the latest technology and startup news.