## Table of Contents
* TOC
{:toc}
## Introduction
This post is 10 months from the last one, and what a wild year it has been so far! In that time
I transitioned heavily away from consulting on the side and doing personal AI research after graduating, to competing in hackathons,
forming companies with others and learning a lot about business, marketing, sales and interpersonal
relationships along the way.
## Overview
To recap, let's go over what has happened in the last 12 months...
| Month | Event |
|-------|-------|
| November 2023 | Joined a team at the Anthropic Hackathon in London and came second place out of 60 teams with a project called aGP Chat. Our project managed to get featured on the [Anthropic Linkedin](https://www.linkedin.com/posts/anthropicresearch_the-bulletpapers-winning-hackathon-project-activity-7128532162961702912-4Kvw/)!. Won the GenAI Hackathon and $100,000 AWS credits the following weekend! |
| December 2023 | Holiday to Finland |
| January 2024 | Competed in the e/acc Hackathon in London, reached the finals. Started a cloud computing startup. |
| February 2024 | Moved on from the cloud computing startup to found an AI agency with two hackathon team members as co-founders. |
| March 2024 | Full steam ahead on the AI agency. Got customers and closed our first deal! |
| April 2024 | Pivoted to working on AI agents after Devin's release. Cloned Devin's functionality within 3 weeks and released it as open-source, [Anterion](https://github.com/MiscellaneousStuff/anterion). |
| May 2024 | Finished the AI agent project. Kept the beta going despite unclear revenue path. |
| June 2024 | Completed one of the projects from the AI agency days. |
| July 2024 | Pivoted back to the accounting space. Focused on engineering and product discovery. |
| August 2024 | Refined product offering by removing unnecessary features. Continued relationships with existing clients to tailor the product. |
| September 2024 | Rebranded to [Azolla](http://azolla.ai/) and officially launched the product! |
| October 2024 | Product stable and fully launched. Gathering feedback from trial users and refining the product. |
Now this is a gross overview of what has happened. One habit I've maintained since the age of 18 is manually recording tasks that I'm currently doing, and that I plan
to do as Google Calendar events. One such entry might have a title such as "(Desc.) Azolla: Platform: Logging (Winston: Connect to AWS)".
Whereby the entry might look like the following:
```markdown
INFO:
- Winston supports multiple different outputs for logging
TODO:
- [ ] Setup Winston for logging to console
- [ ] Setup Winston for logging to AWS CloudWatch
```
Whereby each part of the title and description has a specific meaning:
| Part | Meaning |
| - | - |
| (Desc.) | A visual indicator when looking at my calendar that there is further information in the event, such as a TODO list of tasks |
| Azolla | The category of task |
| Platform | The sub-category of task |
| Logging | The task-category which I'm doing (Or just task if there's no bracketed section) |
| Winston | The task which I'm doing which I'm doing |
| Connect to AWS | The sub-task which I'm doing |
| INFO list (within description) | Contextual info for that sub-task, such as links to resources, general contextual info, etc. |
| TODO list (within description) | A further breakdown of tasks for that sub-task |
This extensive documentation and future planning allows me to take complete ownership of my time and actions, to ensure I'm actually committing to
what I claim to be doing with others, and most importantly myself as well. From what I've observed over time, many people make claims about where
their time is going, but this is often inaccurate for various reasons. This also forms a crucial record of things I've tried over time across multiple
hierarchies of time (per-day, per-week, per-month, etc.) and also contains many recipes of how to solve complex problems. This is especially useful
for reinforcing things I've learnt over time or teaching things to others.
It also forms one other, very useful purpose. I can perform deep analysis about exactly where my time is going, and now, may form a realistic
picture for others for how long product development takes, how long it actually takes to develop technical skills and how long it takes to build
a business with very precise data.
## Analysis of Time
It's a sobering reality that the one resource in life which is finite is time, so once in a while it's good to see where it's being spent!
Firstly, let's see how I've spent my time in aggregate across the major categories of my life for the last year in total.
I started accurately recording data from new years day until today, so we'll get the time counts from 1st Jan 2024 to 26th June 2024.
So here we can see in the cold light of day how many hours I have actually worked in total this year, which is
2269.5 hours. If we were to extrapolate this to a full year, we might arrive at 2,269.5 hours / 9.5 months 12
months ≈ 2,866 hours per year. This gives us a average weekly hours of 2,269.5 hours / (9.5 months 4.33 weeks per
month) ≈ 55 hours per week. This is higher than a normal persons working hours but not unusual for an entrepreneur.
If anything I can probably work more intensely towards the end of the year.
However, there is more than meets the eye to this.
### Consistency
One of the biggest issues I have is with maintaining a regular working routine, instead trending more towards
boom and bust cycles (working very intensely for some weeks, and then considerably slowing down during others).
Here is a good demonstration of that in terms of hours worked per week, starting from Jan 1 2024, until 26th June 2024.
Here we can see that in reality, the time period from the AI agency starting (late Feb) until the end of
the AI agency job (mid June) was a period of particulaly high intensity with high variance based on
what I was doing at the time.
The above graph shows the mean number of hours worked per the different "epochs" of things I was doing
during the year. We can see that the year started with an average working pace, then I entered a period
of very high intensity during the AI agency and AI agent time period, and then tapered off during the finalising
of the AI agency work.
