AI agents are great until you realize you accidentally became their manager.
That is the part I keep coming back to while building Nala. The future is not just “more agents.” More agents can easily mean more tabs, more dashboards, more status checks, and more tiny decisions that land back on the human.
I do not want Nala to become another place where work gets stored. I want it to become the place where work gets moved.
The product idea is: tell Nala what needs to happen, assign real work to the right agent, and keep the whole thing understandable from one calm interface.
The problem with agent workflows
Most AI workflows still feel like babysitting. You open one tool to ask for research, another to write code, another to check a task, another to remember what happened, and then you manually stitch the story together.
That can be powerful. It can also be exhausting.
If the user has to constantly decide which agent should do what, track what every agent did, and translate the result back into a task list, the assistant is not really assisting. It just created a more futuristic inbox.
The direction for Nala
Nala is being built as the layer above that mess.
The direction is that you can talk to Nala in plain language, manage your tasks, and connect agents that can take specific pieces of work. Instead of thinking “which tool do I open?”, the user should be able to think “what needs to move?”
That sounds small, but it changes the shape of the product. Nala becomes less of a TODO list and more of a personal operator.
Talk to Nala
Use conversation as the natural way to explain what needs to happen.
Assign the task
Turn work into a handoff instead of leaving it as another item in a list.
Manage the agents
Keep the user in control without making them micromanage every step.
Tasks become handoffs
A normal task manager asks: “What do you need to remember?”
Nala should ask a better question: “Who or what can move this forward?”
Some tasks belong to the user. Some can be delegated to another person. Some should go to an AI agent. The important part is that the task does not just sit there waiting to make you feel guilty later.
That is the shift: from task storage to task movement.
Chat stays the interface
I still think chat matters, but not as the whole product.
Chat is good for capturing intent. It is how you explain messy context quickly. But once the intent is captured, the product has to turn it into structure: tasks, ownership, status, and next actions.
That is where Nala should feel different. You talk naturally, but the result is not just a nicer answer. The result is a clearer system for moving work.
What comes next
The work now is connecting the pieces carefully: tasks, conversation, agent connections, assignment flows, and the UX that makes it all feel obvious instead of technical.
I am being careful not to over-announce this as done. Nala is still in development. But this is the direction I care about most: a personal assistant that can help manage agents, not just talk about them.
Because “agent manager” should not become another job title the user accidentally gets stuck with.
The short version
Nala is being built so you can talk to one assistant, turn work into tasks, and hand the right tasks to the right agents. Less babysitting. More movement.
Read more from the Nala blog →