Using ChatGPT as a database for managing to-do lists.

Listen to this post

I’m a big fan of using AI to increase my productivity and I really like to explore new ways to use the technology that we have available. That’s why I’ve decided to start using ChatGPT’s memory as a database that I can interface with by voice in order to retrieve tasks for to-do lists.

How ChatGPT’s Memory Works

ChatGPT’s memory allows it to retain information across sessions. These are stored in little snippets that it seems to have access to instantly unlike storing “Knowledge Documents” that it has to always search through. I’m not entirely sure how it is implemented technically, which is why I’m interested in experimenting with its capabilities beyond simply remembering context like “Chris is a web developer”, maybe you can ask ChatGPT to remember groups of items, and ask it to forget things too when they are no longer needed.

Why Use ChatGPT for To-Do Lists?

Traditional to-do list apps and methods often need typing, note taking etc… but as a user of ChatGPT via voice and conversation, I can just ask it to remember things or add/remove things from specific lists. As a software engineer juggling multiple projects, both professional and personal, I need a system that can adapt to my workflow seamlessly. A non-GUI might actually be the solution to offloading my own memory quickly.

How I Use ChatGPT’s Memory

I’ve started using ChatGPT to remember separate lists for my blog ideas and personal tasks. For example, I can say, “Here are some blog ideas, can you remember each of these topics?” Then, in a separate session, I can ask, “What’s on my blog topics list?” or “What’s on my personal to-do list?” and ChatGPT can recall these for me. This makes it feel like I’ve delegated the task of remembering to a real-life assistant. I’ve then come back to it later and asked it to give me a rough plan for my upcoming week. It integrated priority items from my to-do lists like visit my bank and write a specific blog post on my list, allocated time for working on client projects, and even suggested I go Kayaking on the weekend LOL! So not only did it take my to do lists into account, it gave me a personalised leasure activity based on what it knows…. Scary I know…. I’m probably likely to delete some of these personal things I originally let it know like “I like Kayaking” and leave more practical things in. (You can manage what it remembers or clear its memory entirely within the settings).

Software I currently use for to-do lists

For my day-to-day tasks, I’ve been using Trello, which works well for organizing immediate tasks. I keep it as a pinned tab in my Chrome browser, allowing me to quickly move cards around for today, yesterday, this week, and general to-dos. However, Trello is more suited for project management and collaboration rather than simple to-do lists and “remember this for me” type situations. Its an over-engineered solution for what I actually need it for.

A More Human Way of Interfacing with Data

Using ChatGPT’s memory offers a more natural and human way of interfacing with data. I don’t need to remember syntax or navigate through an interface. Instead, I can just talk to ChatGPT, which feels like having a conversation with an assistant who remembers my tasks and helps me manage them.

I can definitely see the applications for things like this and while it doesn’t solve everything, it is definitely the direction a lot of technology will be headed. A smart voice interface that can store and recall data whilst following and remembering instructions may mean that programming becomes a conversation a human and an AI to get a result like this intangible “To-do List App” i have inadvertently created. I mean…. it was already there…. But what other applications could we have create? We could tell it daily how many KM we have run and times, then at the end of the month get statistics on our running habits. The potential is huge!

An Ongoing Experiment

As this is a new concept for me, the true test will be how well it holds up over time. I’m interested to see how the memory feature handles multiple lists and whether there are any limitations to be aware of, like running out of storage or accidental overwriting etc…. As part of this experiment, I’ll be monitoring its performance and will follow up in the future with another blog post detailing my findings.