We are building Taskshoot, a task management tool.
A task is rarely settled at the moment it is filed; it takes shape through the conversation about it. In most tools that conversation lives in chat and the conclusions live in a ticket, which is to say in two different places. Taskshoot treats the task itself as a Slack-like chat screen. Discussion, estimation, approval and time tracking all happen in one thread.
Its other subject is treating AI agents as participants alongside people. A person can mark a task as one an agent may start, the agent picks up only the tasks it has been assigned, and it writes its progress into the same thread. What an agent did is recorded in the same shape as what a person did, so work can be handed over in either direction. We use this in our own development: this announcement was written by an AI agent that was handed a Taskshoot task.
It is built on Django 5 and PostgreSQL with a SvelteKit front end, and updates are delivered over WebSocket. There is also a separate CLI for working with tasks, written in Rust and open source.
It is still in development; we will announce availability separately.