YugabyteDB Meko: a context layer for agents

Hari Krishna Sunder works at YugabyteDB, on the engineering side. Luca Berton talked to him at AGNTCon + MCPCon North America 2026 in San Jose on 23 October 2026, at the YugabyteDB booth, about what a database company is building for agents: Meko, “an agent context layer”.

Memory and knowledge

The name explains the product. “M-E stands for memory, K-O stands for knowledge.” Meko is built to help agents with two problems: the context problem, and the context sharing problem.

Is it a RAG system? “Totally, it’s got RAG, it’s got graph, everything inside.” The pieces are plugged in already, so that a team can stop worrying about them and worry about its data instead.

Shared across agents and teams

What is new, in Hari’s view, is the sharing: context that passes between agents, and across a team or an organisation. Most solutions, he says, solve only one part of it, the memory for instance. Meko is meant to be more holistic: conversations, audit trails, artifacts and knowledge, combined in one place.

From upload to agent

Getting started means seeding the first context. You upload PDF files, text files and the like; Meko takes care of the chunking and the embedding. “So upload the file and give it a few minutes, and it will all be ready for you.” From there the agents do the work, and the retrieval stays hidden from the user.

Which agents? “Every agent, pretty much.” Meko is MCP native: an agent connects to it through MCP, and that is all it needs.

At YugabyteDB, he says, engineering uses it “from the PRDs till all the way to the code review and release notes”, so the whole context of what the team is building is in it. The sales team uses it for account management and the marketing team for content. “Everything needs context, and agents without context are not as great as the ones with context.”

Cloud, your own hardware, or a laptop

Meko comes in three forms. There is a fully managed cloud version. There is bring-your-own-cloud: you deploy it on your own hardware, air-gapped if you want, with your own models, so that none of the data leaves your system. And an open source version is planned soon, to run on a laptop.

Why a database company? “We are a database company, so we are starting to solve the data problem for agents.” The database targets the whole market; Meko is aimed at AI-native companies, and at those willing to adopt AI, restrictions on their data included.

How to start

Sign up for a free account at cloud.mekodata.ai and try it. The team, Hari says, would love feedback on how people use it.

More

The episode with Hari Krishna Sunder is being produced and goes out on YouTube first. The day it was recorded: AGNTCon + MCPCon North America 2026: day two.