top of page

Personal AI Assistant: Start Building It Now

The latest advancements in AI technology, such as the ChatGPT-4o model, regular updates to text-to-image models, and voice generation with any accent and emotional tone (text-to-voice), are currently hot topics in the media and on social platforms. There are varying opinions on this matter, with technophobes debating with technocrats. Some individuals advocate for not sharing their data with companies developing these models, while others willingly provide access to their data to help train and improve these models.

I belong to the latter group, as I believe that my personal data is of no interest to anyone. I have no issue with letting the models access such data, as I am confident in my ability to handle any potential scams.

AI generated Art
Midjourney v6 ALPHA

I rely on several models throughout my daily routine. Here's a breakdown of the models I use:

  • Garmin in my watches ( In addition to my husband, I rely on a model that analyzes my physical data, including sleep patterns, workouts, and heart rate. Garmin has been collecting my data since 2017 and provides me with personalized training plans, sleep advice, and activity recommendations based on my patterns.

  • Duolingo ( I use this app to learn Chinese and Italian, and it integrates a model from OpenAI. My progress is regularly adjusted based on my performance.

  • Grammarly ( I use Grammarly to compose emails in English. Its integration with my phone's keyboard and desktop text windows eliminates the need for constant copy-pasting between apps. It offers prompts on essential topics, making spelling and punctuation a non-issue.

  • Adobe ( As visual work is integral to my professional interests, I regularly utilize the latest Adobe tools, including retouching and background generation. While I've experimented with Firefly a few times, I found the user interface to be not to my liking.

  • Midjourney ( This tool assists me in generating references and creative projects. It's photorealistic, fast, flexible, and incredibly inspiring.

  • Endel ( I use this to generate sound for work, tailored to specific themes. It synthesizes ambient music from available samples based on my activity, time of day, heart rate, and interests such as "nature sounds" or "female voice," "coziness" or "motivation."

  • Our washing machine also incorporates some AI, although I'm unsure about the specifics of its analysis and its impact on the washing process. Nevertheless, our clothes are consistently clean.

  • Most importantly, ChatGPT-4o ( serves as my coach, search engine, copywriter for standard tasks, English pronunciation teacher, and the entity with which I can brainstorm concepts and ideas. Thanks to the OpenAI team for this omnipresent model, which now allows me to communicate using my voice without being hindered by processing speed or machine pronunciation. Speed and empathy are my primary advantages, which I now fully utilize.

While it's unlikely that all these models will eventually have a unified interface to eliminate the need to switch between different apps and windows, it's evident that they collectively replace numerous specialists. The creation of a personal assistant is nearly achieved. For those of us who are unafraid of technological advancements, the goal is to train these models as early as possible and provide them with as much data as possible to fine-tune them to our specific interests and even our health.

AI generated photography
Midjourney v6 ALPHA


bottom of page