Mira Team: From One Assistant to a Team of Agents

I recently decided to go back to school and start a new degree. That means balancing work, studying, training, and everything else life throws at me.

That pushed me to rethink how I use personal AI agents.

From Nineth to a team

A few days ago I shared Nineth, my personal assistant. This grows from there, just a little bit.

Instead of building isolated bots, I'm building a team: one agent for my calendar, one for academics, one for tasks, one for training, and a main agent coordinating everything.

For this new chapter the name is Mira Team 🌸

Who's on the team

Diagram of the Mira personal agents team: Luis proposes and approves, talks to each agent in direct chats or in a study group, and each agent connects to its own source: calendars, the academic platform, a task list, or a training plan

Mira is the main agent, the one coordinating everything. The 08:00 briefing in the corner of the diagram is just an example meeting.

Then there are four agents, each one tied to its own source:

calendar-observer looks at my calendars for free spaces and events, and can create or move events. It works with real data.
platform-reader reads my academic platform: courses, deadlines, grades, and announcements. Also real data.
task-prioritizer prioritizes with a macro view and my weekly plan, and keeps my task list synced as a Q1 to Q4 matrix.
Hy (coach-entreno in the diagram) tracks my progress in strength, cardio, and conditioning, connected to my training plan with its workouts and sessions.

Two ways to talk to them

Direct chats. Each agent is an independent contact. I write, they write back, like a person.

Study group. task-prioritizer, platform-reader, and calendar-observer share a group where they work together. task-prioritizer handles the coordination, platform-reader brings the deadlines, and calendar-observer finds the free spaces.

Hy stays out of that group. It only does training, no studying.

What planning a week looks like

I say "do this next week", and task-prioritizer mentions its teammates.
They publish the plan and wait for my OK. Then calendar-observer creates the blocks on my calendar.

I propose and approve. The agents do the coordinating.

Why a team

The goal isn't to have more AI. It's to have AI that works together, understands context, and helps turn a messy week into a coordinated one.

This is the architecture I'm building around my own needs, and I'm genuinely excited to see how far it can go.

Thanks for reading!