So, not a specific personal case where AI had to negotiate itself more quickly on a product decision. Not faster in itself, but it rather helped me to pass ideas to be able to implement tools myself and help the company by automating processes with software simply by being able to write the software using AI, which was not possible for me before the previous one, you know, the leg. The previous flow was: I have a ticket that the team requested, it is assigned to a member of the development team, and this involves the typical work of work, work, then to go online. But now, in some cases, some of these smaller tasks and faster victories can simply attribute me directly, where I am able to implement it over time a little faster than the typical development cycle simply because I am the only one. I test it, I write it, I test it, I deploy it on its own, so it is a much shorter term. The execution time tends to be shorter for most things, so it was useful in this way.
The tools that helped were, like, yes, Chatgpt, Claude and also the cursor. So, Chatgpt and Claude for many project planning and writing documentation, and in the creation of tickets and stories and epics and documentation, product documentation, while the cursor was the reference tool to write the software and automate things such as Slack alert pieces and the creation of internal board tables that help the internal attribution of data and Users. So, that's right, we are now at a point where you can perform repetitive tasks that I am mainly talking about my colleagues and understand what their repetitive daily tasks are, and what we can automate and remove their daily things, so they have more free time to work on more creative or more pressing problems. It is therefore simply a question of understanding what this process looks like them, then we can connect software in the middle to try to relieve the part of time and efforts to perform these tasks.
So, at the moment, I am currently also setting up a hubspot link with our database and automating our communication. So when users are registered for the first time, we want to send them by e-mail. When users have not reserved a vehicle, say, 30 days, we want to send them by e-mail and engage with them. Currently, the process for doing so is that we have a member of the marketing team that will go to our data visualization tool or to our API, it will export a list of users, glue it in HubSpot, will select the model, then send it. But now, because of the cursor, I am able to create something internally where I simply link our database directly to our HubSpot API, then automate the process for it so that it does not have to copy and paste information. We just do this automatically for several scenarios.
The main parts of automation with AI, specific to product management, were documentation documents. As I said, write tickets, a product roadmap, exit notes, epic stories, things like that. Very useful. And then I try from, similar to the HubSpot API, to connect it to the Jira API. So, either we use MCP, which is a model context protocol, or write software again to bind to the Jira API, so it becomes easier. I can simply speak at AI, mainly in Cursor or Claude, and be able to interact with Jira and ask him: “Ok, look at the backwards, look at the different priorities of the tickets” and help me plan the next 3 to 5 sprints. Tips like that, do you know?
Yes, this is the goal to which I work, but I do not yet have an implemented solution for that. But it is very, very interesting for me, because then a large part of the planning is taken care of, and a large part of the objective is simply devoted to the creation of good tickets, by making sure to make the creative part of the role. Then, it becomes the conversation with customers and internal members where I act as an entry for Jira and documenting all needs and requirements, then prioritize it. And then Claude, or AI with Jira Access, can do a large part of planning according to my predefined priorities. This can change and we can have a conversation together to decide. So, this will be a very big addition of value, which, then, you know, allows you to spend more time doing more analysis of competitors, just practical time with the product and using it more daily to try to improve it in this way, you know?
And I think, yes, it can certainly replace certain PM functions. As I said, in the future, decision -making can be independent. I think that AI does a very good job of prioritization anyway, but a little human contact is necessary for the moment just because of the conversations that I have with my CEO or my COO. For example, that does not know them, so it is really whether there was a way for the whole business to be able to speak to an AI and that there are requests that arrive, and it is to collect information, and it is capable of thinking of a situation as a whole, to prioritize it, but there is just a human person who disconnects things and saying: “Yes, I agree.” I guess you know, it may be the future. I do not know if it will be completely automatic, but certainly certain aspects, at least things like exit notes and things like roadmaps, roadmaps of products and documentation, I think that it is very good to be automated.
