Organising Projects with ChatGPT on a Walk

On a 45-minute walk one Sunday afternoon, I worked through 13 projects with ChatGPT. By the time I got home, each project had a clearer status, priority and next action. The project pages in Notion had been updated and I knew which work needed scheduling in Continuum.

I had been putting this off because it felt like the sort of exercise that would require several hours sitting at my computer. Instead, I did most of it by voice while walking up the hill behind my house.

The dashboard was only the first step

The previous day, I had been organising the projects I work on for one of my clients. Although the individual tickets were in Continuum, I lacked an overview of what was happening across all the projects.

I worked with Claude to review Confluence, Jira, emails and Continuum using MCP servers, then created a project status dashboard. This gave me a single view of the current position, upcoming work and any blockers across the portfolio.

It immediately reduced the cognitive load for that client, so I created something similar for my own projects. I had a growing collection of products at different stages, including Weather Alert Spain, AI Splicer, Pass PM Exams, Battle Isle, My Portable Life and Continuum itself.

The new dashboard told me that I had 13 projects. It also exposed the real problem. I knew what existed, but I had not decided what should happen next with each one.

Whenever I found some free time, I would pick up whichever project happened to be in front of me. I might fix a bug, adjust a design or start thinking about a new feature. Meanwhile, most of the other projects remained untouched. Each one sat somewhere in the back of my mind, competing for attention.

Taking the session away from my desk

I like to take long walks. I work from home, which I particularly enjoy, but without a commute I can easily get stuck in the house. Walking gets me outside and gives me space to think. As a product manager, I have often found myself fitting puzzle pieces together away from my desk.

More recently, I have also started leaving WhatsApp voice messages for friends while walking. Finding time for a phone call can be difficult when people live in different time zones and have children, so we leave messages that the other person can answer later.

On this particular walk, I finished sending my messages and realised my head was still full of unresolved thoughts about my projects. I looked at the ChatGPT app and realised I could use the same approach.

ChatGPT could access Notion, my emails, GitHub and Continuum through MCP servers. These connections meant it could pull up the information already stored across the different systems and update it as we worked. I opened the project list and started talking.

Working through the projects

We took the projects one at a time. ChatGPT pulled up the existing project page and, where useful, checked the backlog or GitHub repository. I explained what had happened since the information was last updated, what I thought was still outstanding and how much attention I realistically wanted to give it.

We then agreed the current status, priority and next action before moving to the next project.

For Weather Alert Spain, the product was largely finished and going through its migration process. The next useful work was reviewing the code and completing the cutover rather than adding more features.

Pass PM Exams became the next product to migrate after AI Splicer. Its main blocker was no longer the application itself, but creating a proper set of exam questions and finding people to test them.

Continuum had reached a suitable MVP. I could keep refining it while using it, but its next meaningful stage was testing it with other people and seeing whether it solved the same problem for them.

Battle Isle had also reached its MVP. The two-player game was working and I was no longer finding many bugs. I had spent a lot of time adjusting menus and other details, so I made it a lower-priority project. Updating the graphics would be the next deliberate piece of work, but until then I would stop tinkering and only fix bugs as they appeared.

My Portable Life remained active. We also added the eventual move from my old WordPress travel site to the backlog, including the redirects that would be needed when the new site launched.

Other projects were placed on hold, given a planning session or left as ongoing work. Some needed a specific date to return to them. Others simply needed an honest decision that they were currently good enough.

Why voice helped

There were around five minutes between my voice messages while ChatGPT reviewed the information and responded. That gave me time to walk and think before moving on to the next project.

Speaking also changed the way I approached the review. Sitting at my desk, I would have been tempted to open a repository or start fixing something. On the walk, I had enough distance to think about whether the project deserved attention at all.

I would normally say that going for a walk should be a chance to switch off. Turning every spare moment into work is hardly a healthy ambition. On this occasion, though, the projects were already taking up space in my head. Working through them reduced that pressure. Each conversation ended with a decision and somewhere for the thought to go.

Connected AI as a product management tool

The useful part of this exercise was the combination of voice and access to live project information. A normal voice note would have captured my thoughts, but I would still have needed to sit down later, check each system and turn those thoughts into actions.

Here, ChatGPT could compare what I was saying with the information in Notion, GitHub and Continuum, ask questions where something did not match and record the decisions as we went.

The dashboard gave me visibility across the projects. The conversation helped me make the decisions the dashboard could not make for me.

By the time I got home, none of the projects had magically been completed. I did, however, know what each one needed, which ones mattered now and which ones could wait. They no longer all needed my attention at the same time.