ekoDB Beta is live! Try it for free today!
The agent runtime
AI agents that live inside your database
Most AI stacks bolt an agent onto a database over the network and then spend the rest of their life keeping a second system in sync. ekoDB puts the agent runtime inside the deployment, so your agents read and write your collections directly, with native retrieval-augmented generation and vector search built in.
Give each agent a system prompt, scoped tool permissions, goals, and scheduled tasks, then run it from the dashboard, your terminal, or embedded in your product. Bring your own OpenAI, Anthropic, or Google Gemini key and it never leaves the server.
The agent runtime
Your AI agents live next to your data
Every deployment can run an agent server. Host AI agents with their own tools, goals, scheduled tasks, and memory, all talking directly to your database from the dashboard or your terminal. Run them for your own team, or embed them in your product so your customers can chat with them too.
Chat with your data
Ask questions in plain English. Your agents answer straight from your own collections using ekoDB's native RAG and vector search. There's no separate pipeline to build and no second database to keep in sync.
Agents with goals & tasks
Give each agent a system prompt, scoped tool permissions, goals, and scheduled tasks that run on their own. These are real workflows, not just one-off prompts.
Tools & templates
File operations, web search, HTTP fetch, persistent memory, and reusable templates all come built in. Bring your own OpenAI, Claude, or Gemini key and it stays server-side, never exposed to the client.
It tunes and debugs itself
The agent looks after what it builds. It explains slow queries, spots a missing index and creates it, surfaces which data is hot, and reads the server logs to find problems. The database tuning that usually needs a specialist happens in the conversation.
What you can build
From a chat box to a workflow that runs itself
Customer-facing copilots
Embed an agent in your product so your users can ask questions and take actions against their own data, with your system prompt and your scoped permissions deciding what it can touch.
Internal operations
Give your team an agent that answers from your live collections, runs reports on a schedule, and handles the routine data work that usually turns into a backlog of one-off scripts.
Retrieval and RAG
Agents answer from your collections using native vector search and full-text search, so retrieval happens in the same system that stores the data instead of a separate pipeline you have to keep in sync.
Autonomous workflows
Goals and scheduled tasks let an agent do recurring work on its own, from nightly summaries to data cleanup, without anyone kicking off a prompt each time.
Frequently asked questions
Which models can I use?
Bring your own OpenAI, Anthropic, or Google Gemini key. The key is stored server-side and is never exposed to the browser.
Do agents remember context between messages?
Yes. Each agent has persistent memory along with a system prompt, scoped tool permissions, goals, and scheduled tasks that can run on their own.
How do agents reach my data?
Directly. They query your collections through ekoDB's native retrieval-augmented generation and vector search, so there is no separate retrieval pipeline and no second database to keep in sync.
Can my customers talk to them?
Yes. You can run agents for your own team, or embed them in your product so your customers can chat with them too.
What tools do agents have built in?
File operations, web search, HTTP fetch, persistent memory, and reusable templates all ship with the runtime.