Launched this week
Clawther
Give your OpenClaw agent a real task board
456 followers
Give your OpenClaw agent a real task board
456 followers
I accidentally connected my OpenClaw agent to Notion and realized something: Chat is the wrong interface for managing an agent. 👎 Agents need a task board, not a conversation. With a task board you can track multiple tasks in parallel and share the power with your team just like you would collaborate with a human teammate. ✅
oh this is so much needed, especially in the context of governance issues across ai agent entreprise use that's been spawning lately.
Clawther
@alexandru_hirsu That’s exactly the problem we’re trying to solve with Clawther. Curious to hear how you’re currently managing agents on your side.
Hey everyone 👋
This product started almost by accident.
For the past few weeks I’ve been experimenting a lot with OpenClaw agents. Like most people, I was interacting with them through chat.
And it worked… until it didn’t.
Very quickly the conversation became messy: tasks buried in messages, things repeated, no visibility into what had already been done.
One day I accidentally connected the agent to a Notion task board.
Suddenly everything felt different.
Instead of chatting with the agent endlessly, I could just create tasks.
The agent would pick them up and execute them.
And my team could see what was happening.
That’s when it clicked for me.
Chat is the wrong interface for managing agents.
Because of ChatGPT, many of us started thinking that chat is the interface for AI. That made sense when AI was mostly answering questions and giving information.
But now AI can actually do tasks.
And when work happens, conversations become chaos.
I believe we’ll see a shift:
from chat interfaces to task interfaces for agents.
Agents don’t belong in conversations.
They belong in task boards.
So I built Clawther: a task board layer for OpenClaw agents.
With it you can:
• manage multiple tasks in parallel
• track what the agent did
• collaborate with your team
• assign tasks to different agents
It’s still very much an MVP, but it’s already extremely useful for the way we work.
Curious to hear how others are managing their agents today.
Happy to answer any questions and hear feedback 🙏
Clawther
@bengeekly 🚀👏🏻
@billchirico Great question!! Thank you so much for the support! The main thing we’re trying to rethink is the interface where agents actually operate. Most current tools either keep everything inside chat or build dashboards that are still basically prompt interfaces.
So instead of asking an agent to “do something” in a thread, the system revolves around a task board where agents pick up, update, and complete tasks. The goal is to make execution visible: what exists, what’s running, what’s blocked, and what’s done.
For assignment and delegation, there are a couple approaches. You can manually assign tasks if you want explicit control, but the system can also route tasks automatically depending on the type of task or the agent responsible for that capability. For example, a research agent might pick up research tasks, while another handles writing or automation.
In terms of ordering, the board structure itself helps a lot. Tasks can have dependencies, priorities, or stages (to-do → doing → done), which lets agents understand what should happen next instead of just responding to prompts sequentially.
The big idea is turning agents from something you talk to into something you can actually coordinate work with.
And thanks again for the support on the launch! 🚀
@rania_rimali How do you guys handle auto routing of tasks? How does a research agent know to pick something up instead of a developer, for instance? Curious about how this works.
100% agree that chat is the wrong interface for agents doing real work. We've been building project management tools and the same thing comes up constantly. When agents are just responding in a thread, nobody knows what's actually getting done vs what's still pending.
The Notion task board angle is clever. Are you planning to support other PM tools eventually or going all-in on Notion as the backbone?
@mihir_kanzariya We are focusing on Notion first, and likely on building our own interface as well to push what’s possible and define what a Kanban board needs to make human–agent collaboration truly powerful. For example, for humans, the core columns are usually just Backlog, To Do, In Progress, and Done. With agents involved, we’ve added two more stages: To be verified by the agent and To be verified by a human.
We’ll add support for other project-management tools only after we clearly identify our killer features. Once we know what truly differentiates us, we can make focused decisions and build the system with a clear direction.
Clawther
Totally agree @mihir_kanzariya ! Once agents start doing real work, a chat quickly becomes messy and it's hard to see what’s actually done.
Product Hunt
Congrats on the launch!
What are the clear benefits of giving your agent a taskboard and does it only work with OpenClaw? I'm curious how you see Clawther differ from @VidClaw .
@gabe Thank you.
A task board lets you schedule multiple tasks, go to sleep, and wake up with things already done.
If you manage agents through chat, you usually have to stay in front of the conversation to guide every iteration yourself. My own experience with OpenClaw in the early days had me waking up and, with my eyes half open, answering small questions in Telegram just to keep it running.
The difference comes from the structure.
A task board lets you organize and run multiple tasks in parallel. And that opens many opportunities.
With Clawther, an agent doesn’t just execute the task, it also checks the result, suggests improvements, and can iterate up to 6 times while you sleep. Then move it to “To be validated by human.”
Clawther is currently built on Notion, which is a big advantage if your team already uses it. It allows human + AI collaboration around tasks instead of isolated conversations.
Both Clawther and VidClaw are still very early, so what I’m describing today will probably evolve a lot in the coming weeks.
But our hypothesis is simple:
As AI becomes agentic, we’ll need interfaces that go beyond chat.
From what we’re seeing so far, a Kanban-style task board feels much more efficient.
Now the real question is:
What rules and killer features should a Kanban redesign get to have another ChatGPT moments for AI?
Interesting idea. One quick question: How does Clawther synchronize task state with the OpenClaw agent?
For example, when the agent completes or updates a task, does it push updates through a specific API/webhook, or does Clawther poll the agent’s state periodically?
@konstantinalikhanov When a task is completed or there’s progress on it, OpenClaw updates the task status and posts a comment via the API. Then, using cron jobs and the Notion API, it periodically re-checks the task, tests the current result, and continues iterating until it determines the task is truly done
@basma_el_khamlichi That’s a great question!! One thing we’re already thinking about based on early feedback is better task prioritization and management. Since the goal is to treat agents more like teammates working from a task board, we’re exploring ways to make it easier to organize tasks and track what the agent is working on at any given moment.
@basma_el_khamlichi Multi agent: Ability to use different agents per task, I already have it configured and testing.
Collaboration: I sometimes ask Agents to create a task for other agents, but it's basic and linear.
Model per task: Will need to have a proper dashboard and move away from Notion
Do you have any ideas to add ?
Clawther
@basma_el_khamlichi Thank you! We’d love to hear your feedback so we can improve and optimize the next features.