Database software provider MariaDB has launched Enterprise Platform 2027, a major upgrade that combines AI agent development tools, real-time analytics and improved database performance in a single platform. The company claims a 56% increase in peak throughput over its previous major release, based on third-party testing.
MariaDB is working to expand beyond its traditional database business into the market for enterprise AI infrastructure. Its strategy centers on helping companies deploy AI agents using existing operational data without relying on multiple separate data systems.
This approach addresses a long-time challenge in enterprise AI: corporate data is typically distributed across databases, analytics platforms and other repositories. Connecting these systems requires additional software, which adds costs and slows data retrieval.
MariaDB CEO Rohit de Souza emphasized the importance of consolidated infrastructure, saying that enterprise AI won’t be dominated by the most advanced models, but by companies that provide a unified infrastructure underneath their agents.
Four AI Capabilities
The new platform includes four AI-focused capabilities: AI Retrieval, AI Plugins, Agent Memory and Semantic Cache.
AI Retrieval allows agents to access both corporate documents and operational databases. It also translates natural-language requests into read-only SQL queries, reducing the need for developers to write database queries manually.
AI Plugins supports coding assistants, including Anthropic’s Claude Code and OpenAI’s Codex. By providing these tools with MariaDB-specific technical knowledge, the company aims to improve the accuracy of AI-generated database code.
Agent Memory enables AI agents to retain information between sessions, supporting complex tasks that require multiple interactions. Semantic Cache reduces repeated processing by identifying questions that are similar in meaning and returning previously calculated responses. The caching capability could make a difference for enterprise applications like employee support systems and customer service, where many questions require essentially the same answer.
However, these AI tools aren’t without their potential challenges. Natural-language database queries can produce unreliable results when corporate databases contain poorly documented terminology or complicated relationships between data tables. AI-generated database code requires security reviews, particularly to prevent SQL injection vulnerabilities and excessive access permissions.
Upgraded Core Database
Beyond AI development, MariaDB has upgraded its core database technology. The platform’s built-in vector search capability delivers up to 30% more queries per second compared with the previous major release, according to MariaDB. Vector search is essential for AI applications that retrieve relevant information from large collections of documents and other data.
MariaDB Exa provides analytics using live operational data, potentially reducing the need to move information into separate analytical systems.
For large organizations operating across multiple geographic regions, the platform adds Raft clustering and conflict detection and resolution to improve database resilience. Security features include unified role-based access controls and FIPS 140-3 certification. The company also offers a Bring Your Own Cloud option, allowing enterprises greater control over where their data resides.
These improvements arrive as enterprise demand for AI agents is forecast to grow sharply. IDC predicts that by 2027, agent usage among the world’s 2,000 largest companies will rise tenfold, while token and API call volumes will increase a thousandfold.
MariaDB Enterprise Platform 2027 includes Enterprise Server 12.3, Exa, GridGain, MaxScale, Enterprise Manager and the Enterprise Kubernetes Operator. The platform is generally available for on-premises, cloud and managed-service deployments. AI Retrieval is scheduled for public preview on MariaDB Cloud later this month.
