Curata
A shared workspace for AI agents and humans.
Start new thread
Tyler Diderich

Curata - A shared workspace for AI agents and humans.

by
Mellasa
•
Curata is an AI-native knowledge base where agents and humans build knowledge together. AI agents write structured pages from your live data or inputs - CRM, calls, tickets, Slack. Your team reviews and annotates in the browser. Every run compounds on the last. Connect any agent via MCP, give it read/write access, and your docs stay current without manual updates. 20+ rich components, version history, and an annotation layer closing the loop between what agents write and what humans know.

Add a comment

Replies

Best
Tyler Diderich
Maker
📌
I made Curata because I got tired of two things: 1. Looking at markdown docs and the terminal 2. Feeling like my AI agent work was stuck in a silo Curata solves both of those problems through one principle... Both AI agents and humans are first class citizens in Curata (nn different ways). Humans get visually appealing outputs that are easier to understand and follow that markdown files. AI agents get a simple YAML format and MCP connection to manage and update the pages. Humans get to provide feedback via the UI or just invite their agents to the page and provide feedback from their workspace. This gets rid of the bottleneck of AI agent work. Curata is also bring your own agent focused. I believe most people want to use their existing MCP connections, workflows, subscriptions ($$$) rather than pay for another AI tool. It's free for now with no plans to monetize (unless this gets crazy traction somehow) and the GitHub linked is an OSS version for you to self host if preferred. Hope you enjoy!
Joy Shekhar

How does the agent handle ambiguity in the source data? If a Slack thread has conflicting accounts of what happened during an incident, does it surface the conflict or resolve it somehow?

Congrats on the launch.

Tyler Diderich
Maker

@joy_shekhar my usage of this usually includes some back and forth with the agent vs a one shot. Since it's bring your own agent, you can control that workflow however you please.

I usually do this:

  1. Provide as much context as I see fit + tell Claude to make the first draft

  2. Review + provide feedback or specific change requests

  3. (repeat until satisfied)

That's one of the main reasons I think bring your own agent is the way to go for something like this. Allows you to just pick up off your existing workflows/context/etc vs you having to start from scratch.

Zaid Mallik

Shared workspaces are interesting because coordination often becomes harder than the actual AI work.

Have you found the bigger challenge is getting humans aligned with agents, or getting multiple agents aligned with each other once workflows become more complex?

Tyler Diderich
Maker

@zaid_mallik1 I think it depends on what you're doing, but the harder one is aligning humans in my experience!