Yugabyte, the distributed AI database expert, is debuting Meko, an agent-native data infrastructure designed specifically for multi-agent AI systems that work and learn together.
According to Yugabyte, as enterprises increasingly deploy AI agents to automate complex workflows, Meko solves a fundamental and growing challenge: how to give agents the persistent, shared memory and knowledge they need to compound their learning over time.
Meko introduces a new storage paradigm that gives AI developers a shared layer for memory, knowledge, conversation history, and observability—replacing the brittle stack of relational databases, vector stores, document stores, caches, and object storage, which IT teams are forced to stitch together.
Built from the ground up to support AI data needs such as knowledge, memory, conversations, and traces, Meko exposes agent-native actions, such as “add knowledge,” that directly represent the AI data constructs used by AI agents.
Developers can now build and interact with these abstracted functions through standard interfaces (MCP) while Meko automatically manages how data is stored, indexed, and optimized across underlying storage systems.
Meko is built on YugabyteDB, a horizontally scalable, PostgreSQL-compatible distributed database that natively supports SQL, NoSQL, vector, time-series, and graph queries. This means a single query can span multiple data models without stitching, unlocking higher performance and lower costs, the company said.
Meko enables agents to share learnings with other agents and humans alike. Context management is complex and requires tremendous engineering effort, even more than application development. Meko simplifies it by unifying data constructs required for modern multi-agentic applications. In addition, it supports collective memory, which aggregates information across the system rather than relying on localized agent memory.
"There is no data infrastructure today that seamlessly allows combined learning and sharing across agents and humans," said Karthik Ranganathan, co-founder and CEO of Yugabyte. "Meko solves this through collective memory, a shared foundation where every agent's learning compounds across the entire system, not just within a single context window."
As an agent-native data infrastructure, Meko enhances the AI data layer and provides three key values to enterprises:
- Shared knowledge and compounding memory: Meko introduces the concept of a Datapack. This portable, multi-tenant data store persists per-agent memory while making knowledge shareable across an entire system of agents. This means that when one agent learns something, it appends the new information to its knowledge, and all users benefit.
- Built for the economics of agentic workloads: Meko is architected from the ground up for the bursty, variable nature of agentic applications. Its serverless, multi-tenant design means costs stay low when agents are idle and scale seamlessly when they're active.
- Auditability for an era of AI regulation: Regulators worldwide are moving toward mandatory documentation requirements for high-risk AI systems, including under the EU AI Act. Meko provides a complete, traceable audit trail of what agents learn, how they share that knowledge, and what data operations underpin every interaction.
By delivering an agent-native architecture that manages knowledge, memory, and conversations, Meko provides developers with the foundational infrastructure they need to build the next generation of intelligent applications, the company said.
“Meko is about removing the friction between ideas and production,” added Ranganathan. “When developers no longer have to worry about how to manage agent state, they can move faster and build more powerful AI experiences.”
Meko is currently available as a fully managed service, allowing organizations to get started quickly without managing infrastructure. The platform will support multi-region and multi-cloud deployments, enabling global scale and high availability for production AI systems.
Additionally, the company plans to make Meko available as open-source software and follow a community-driven development model. Developers can run Meko locally for experimentation or deploy it across private clouds, public clouds, or hybrid environments.
Developers can get started by requesting access at www.mekodata.ai.
For more information about this news, visit www.yugabyte.com.