Fernão: The Personal AI Redefining Task Management

Improvement of the Calendar Function

In the pursuit of enhancing Fernão, my personal AI assistant, I made significant changes to the calendar function. Initially, Fernão used the ICS format to retrieve calendar events, but this method proved to be inefficient. Indeed, the ICS format does not allow for native filtering of events, which forced Fernão to download my entire schedule just to extract a few relevant meetings. It was akin to bringing home an entire library just to read a single sentence in a book.

To solve this problem, I explored other options and discovered that Google offers calendar access via an API that allows for native event filtering. Thanks to this API, Fernão now retrieves only the events it actually needs. This improvement has significantly reduced the schedule generation time, from nearly five minutes to about twenty seconds. With this new pipeline in place, I also refactored the surrounding logic to make the entire function cleaner and faster.

Here’s an example of how Fernão now retrieves calendar events via the API:

def get_events_for_date(target_date=None, use_api=True):
    # Retrieves events for a specific date from Google Calendar.
    # First tries the API (if use_api=True), falls back to ICS if the API fails.
    target_date: datetime.date object for the target day. If None, uses today.
    use_api: If True, first tries the Google Calendar API. If False, uses only ICS.
    # List of dictionaries of events.

New Features

In addition to technical improvements, I added new features to Fernão. Now, in the calendar view, I can mark tasks as complete, and they are automatically synchronized with the Microsoft To-Do app. This integration simplifies task management and could foreshadow a trend where users build their own personal operating systems by integrating various tools.

This makes me think that over time, many of us will end up building our own "personal operating systems," assembling workflows from our favorite tools, integrating what we need, and replacing components whenever better options arise. If features become easy to replicate, loyalty to specific platforms will weaken.

This also raises an interesting question: will companies eventually try to lock down their systems by restricting APIs and external integrations? It’s possible. But isolating themselves is unlikely in the long run. If users cannot connect tools to their own workflows, they will simply go elsewhere (...and will this behavior be limited to advanced users?).

With this new capability in the Schedule Maker, I implemented another feature for Fernão, suggested by one of the readers.

Introduction: the Task Decomposer

Fernão's new role is to decompose tasks. The Task Decomposer follows a simple workflow:

• Start with a large generic task in Microsoft To Do;
• Add context on how the task should be broken down;
• Fernão decomposes the task and builds a plan;
• The resulting tasks are saved in To-Do with due dates and later appear in the daily assistant.

Here’s how the Task Decomposer appears in the sidebar:

Let’s take a concrete example. One of the largest tasks on my To Do list is "Project Documentation at DareData." This is not a small operational task; it’s a structural project.

The goal is to consolidate and formalize the company’s internal knowledge in Notion, ensuring that key questions about the company are clearly answered in our Knowledge Hub. This means reviewing each department, identifying gaps, creating or refining pages, properly structuring information, and assigning clear responsibility for each section.

In practice, this requires audits, writing, and governance decisions. It’s not something you "just do" in one go, and I want to finish this within three weeks. Realistically, I can dedicate between 30 minutes to an hour each day.

This is exactly the kind of task that benefits from decomposition, so I will use some contextual information in the Task Decomposer and transform this mega-task into an action plan:

• I am currently building our Knowledge Hub in Notion. Here’s the current structure of Notion and departments:

  Network Members Manual

  [DareData Network Members Manual]

  Sales & Partnerships

  Marketing (Brand and Gen-OS)

  [Technical Specialists] [Account Managers]

• Review each department and create the pages that make sense (search the web for more context on what DareData is if you need ideas on pages to create for each department or use your knowledge of what absolutely needs to be documented in a 130-person B2B company).

• Ensure that each page has an Owner.

• Review the suggestions recorded by my team in the suggestions; I can probably choose one or two each day.

• Add some departments that are missing: Administration, Key Members, Finance, Product.

After clicking "decompose," Fernão will start forging:

We can now review the generated tasks and plan and submit them to my Microsoft To-Do. I also need to assign which list on Microsoft To-Do they will be allocated to:

Adjustments and Improvements

While testing this, I realized I wanted to adjust two things.

• First, Fernão is supposed to be a medieval chronicler, not a fantasy warrior, so I will change the forge design. His clothing should have a medieval look.

• Second, on a more practical note, I need a "Submit All" button. Submitting tasks one by one is tedious, especially when decomposing larger projects.

Let’s see how this looks in the Microsoft app:

I will probably remove the emoji. This kind of "emojification" is very typical of code generated by LLMs in the front-end, and it really annoys me.

Anyway, building this new feature for Fernão has been really satisfying. If you have ideas for additional features, I am open to suggestions!

Here’s the current prompt I use for task decomposition. I will continue to refine it as I experiment and observe how it works in real use cases.

name: task_breakdown
description: Break down a task into actionable sub-tasks of 20 minutes
max_tokens: 4096

You are a productivity expert helping to break down complex tasks into manageable...