Personametry + ChatGPT = personametry.ai (a truly personalised AI)

Around this time of year with 6 months already past us, I spend some time reflecting on my Personametry and RAGE model - keeping in check how I'm performing against my own personal and professional goals -- just like we do in business with mid-year performance reviews. 

If this is the first time you've come across my work on Personametry, here's a view of my time spent in 2022 compared to 2021, click here. 

My workflow until then was:

1) Export data from Harvest

2) Update Amazon Quicksight dataset by importing the Harvest export, creating transforms

3) Create a new analysis

4) Modify all the charts and views to include the latest data, publish a new dashboard

5) Create a Google slide deck 

NO MORE!! ENTER CHATGPT and the NOTEABLE plugin!

My new workflow is now much more simplified:

1) Export data from Harvest

2) Import to Quicksight, creating transforms

3) Export CSV from Quicksight

4) Prompt ChatGPT to produce the insights in Noteable I would normally have manually created

Okay, so I spent my Sunday locked up in my office playing around with ChatGPT geeking out on data analysis & visualisations. So not much gain in time productivity there - but the learning was fun and immensely rewarding. Yes, a good investment of time, spent learning and preparing to adapt for the new world of AI disruption!

I've been on my data capturing journey since 2015. My end goal to end up with an AI personal assistant that truly understands me. I suspect realising my aspiration isn't that far off and neither far fetched for that matter. My entire workflow would be a fully integrated AI assistant that has the ability to track all my activities, by the minute (imagine "personametry.ai"). To get there, the next simpler step would be to automate the data ingestion piece, leaving me with the only manual entry of starting and ending tasks. My next experiment would be to use the raw data export from Harvest, without doing any data transforms and let the AI do that for me. If that works, I can build in automation that does monthly data imports and produces insights for me automatically. Ideally, I would build personametry.ai as a task into my personal assistant. Imagine a time when we're all wearing a device that "just knows" what we're doing, who we're interacting with, and what we're spending our time on? This device, i.e. "my AI" or "personametry" will act as a guide, coaching us along the way to improve - and hold us accountable - calling us out on what we're paying attention to (i.e. deviating from our goals.)

Lessons Learnt - Still early days but very hopeful

• The tools are still early days, but still nevertheless very powerful and will definitely improve my productivity in future.
• It takes a few attempts to load data files with the plugin, CSV seems to cope much better than XLS files.
• ChatGPT/Noteable uses different methods each time for approaching the data analysis - some coaxing on the nature of the dataset produces better insights.
• I didn't need to edit any code myself so there's an immediate empowerment driver right there.
• Don't trust every output though, as the AI can get things wrong
• Double check calculations, ChatGPT still doesn't seem to get some simple math right the first time round
• It is amazing what one can accomplish with simple, clear prompting
• I am definitely going to learn more
• I had so much fun learning, I was in a state of flow for 10 hours and couldn't stop thinking about the world of possibilities of this technology!

Here's a video of ChatGPT prompts - Play at fastest speed (sorry, no time to edit)

Here's the conversation history with ChatGPT

Round 4

Round 3

Round 2

Round 1

Here's the Noteable project that I've released publicly

Here's a screen grab of the visuals in a nice slide view

Here's all the questions ChatGPT answered in various sessions

First provide insights on the following - remember to give each one a heading in your response and a new section in the Noteable notebook:

1. Stats for nerds - Table showing the number of data points captured for each NormalisedTask by year (basically a count of each NormalisedTask)
2. Stats for nerds - Table showing the total hours logged for each NormalisedTask by year (basically a sum of hours for each NormalisedTask)
3. A bar graph showing for each MetaWorkLife, a bar graph of how time was spent in each year
4. Create a dashboard scorecard for each persona going back 5 years
5. Show a pie chart of work-life balance for 2023 and provide a summary if it’s looking good or bad
6. Show a table for YoY comparison between 2023 and 2022, using indicators (up/down) to show where changes are coming from
7. Show a waterfall graph using the first 6 months of 2022 hours as total start, and show how 2023 hours contributes to increases or decreases across the top level personas “PersonaTier2”
8. Analyse my work hours Year-on-Year for 2022 compared to 2023, month by month, and conclude the percentage increase or decrease of work hours? That is, give me a percentage increase/decrease for 2023 work hours. Include bar side-by-side bar graphs and show a reference line of 168 hours for monthly work hours.
9. Provide a heat map of work hours over all the years. The focal point for heat map work hours should be 168 hours. Values above 168 hours considered hot, values less than 168 hours is cool.
10. Analyse my spirituality over time. How am I tracking? Give me a deep analysis and insights. Which year was the watershed year for me?
11. What insights can you give me about my family time? Go as deep as possible.
12. Tell me how am doing in the social context?  {NormalisedTask}='[Friend] Social','Social’. Use the SocialContext and SocialEntity field as inputs and provide your insights.
13. Please provide me a breakdown of “Me Time”. As an Individual, where am I spending my time in?
14. Provide insights on my health, fitness and well-being  NormalisedTask='[Individual] Health, Fitness & Wellbeing'
15. Show me “real me time” including everything containing “Individual” but excluding sleep time.
16. How am I doing as a Husband?
17. Am I overworking?
18. Which year did I work the most hours?
19. Which year did I work the least hours?
20. Since starting my new job in 2021, which months have seen the most hours?
21. Since 2021, how many weekends did I work (show me for each year)?
22. Since 2021, when do I start logging work hours (“StartedAt), and when do I end my work (“EndedAt)?
23. Show me a table for each year from 2021, my average daily work hours?
24. Compared to first 6 months of 2022, how am I performing in 2023?
25. Which of my personas seen the most improvement in 2023 compared to 2022? Dive deep into the personas using “NormalisedTask” as well.
26. Which of my personas seen the most improvement in since 2015 to date?
27. Am I getting enough sleep?
28. Am I getting enough hours for physical exercise, health and well-being?
29. At which days am I most spiritual?
30. Given my goal at maintaining a healthy work-life balance or harmony, what inisghts can you provide me in this area? Use a benchmark of a 40 hour work-week and 160 hours work-month.
31. How many hours extra did I work each year? 
32. Since I started a new job in 2021, how many hours extra did I work from 2021 to date (last available data point). Show the results in table format that shows only the extra hours above 168 hours (168 is what consultants use as the typical monthly work hours). Assume my billable hourly rate is R2000 per hour. How much income did I lose due to extra work hours?
33. In 2022, my data showed an improvement in my work hours (that is, my work hours must be trending down to meet the recommended 168 hours/month). In 2023, how am I doing? 

Credits to my learning

Here are two videos that kick-started my learning: