So for the Slack bot - we’re a car rental company, so we run into operational issues quite a bit. This involves vehicles getting too far away from the home location or vehicles being low on charge. For example, a few weeks ago, we had a user drive from California all the way to the East Coast - like Chicago, New York, or Pennsylvania. They drove pretty far. They were out of state, and we didn’t catch it. So I ended up creating a Slack bot, connecting our database. Like most software companies - or most companies in general - they’re just CRUD businesses, right? Like, databases with business logic on top. CRUD being Create, Read, Update, Delete. So now, essentially, if you have access to the database, you’re able to very quickly implement things based on business logic, because the database is the main component of a lot of these SaaS companies or modern businesses. Especially with direct database access, I was able to now compare every one hour - basically like a cron job that gets triggered. We go through the database, check every vehicle, check the home location, check how far it is. And if they’re very far - like over 100 miles - we just alert the team, and they know right away that they should be watching it. In this case, an expensive vehicle was thousands of miles away - an expensive asset. They’re saving possibly $5,000 to $10,000 or more in asset value. That’s a big deal. So things like this go a long way in reducing cost, maintaining security - especially in our use case with car rental or car sharing. Another thing I created was an internal dashboard - which is basically a website. First, let’s understand the pain point. We have users, and the company provides electric vehicles as an amenity. Sometimes people refuse to admit they were using the vehicle when something happened - like getting a ticket or toll. These don’t get assigned to the person driving the car at the time - they come back to the company. So every month, we get a list - say $500–600 worth of tolls - and we have to go through the painful manual process: when was the toll, what day, what time, which vehicle, etc. Then the customer service team would spend hours figuring out who it was, notating that, and charging the person. What I automated with this dashboard is: they just upload a CSV file, and a script runs. It compares the file to the database. So now, after getting the toll info - like the car and timestamp - we can find out exactly which car it was (based on the license plate), match it with the booking, and identify who was driving at that time. Then the team can bill those users in bulk. What was once a manual, time-consuming process now takes seconds. For the HubSpot integration - I haven’t fully implemented it yet, but the goal is to automate communication with users. We’re a car rental or car sharing company, so when users sign up, they go through steps like DMV checks. But some users complete only part of the process. Right now, the team goes into Power BI, downloads a list of users who didn’t finish onboarding, and emails them manually. Same for users who signed up but never made a trip, or users who haven’t booked in the past 30 days - we want to engage with them. The current process is: download that list as a CSV from Power BI, upload it into HubSpot, select a template, and send the emails. This happens a couple of times a month. The goal is to replace that with a direct integration between the database and the HubSpot API. Since the database tracks user activity, we can identify users who meet certain criteria and pass that info to HubSpot automatically - daily or weekly. That way, no one has to download/upload anything. Communication campaigns can be automated quickly across multiple use cases. For the Jira API automation, it’s just to make my life easier as a product manager. I want to talk to Claude or Cursor and ask: what Jira tickets do I have, how many in the backlog, how many story points, etc. Either through an MCP (Model Context Protocol) or directly with the Jira API, like we did with HubSpot. I’d set up a dashboard where I can talk to Cursor, linked to Jira or MCP, and get updates on what’s in the sprint, backlog, who’s assigned what, how many tickets or story points are being completed, etc. That helps me plan future sprints and product roadmaps better. This isn’t fully implemented yet, but the MVP shows promise. I’ve had Cursor output CSVs that are easy to read. I can also ask it questions like: “Let’s plan based on the backlog - we have 300 tickets - how do we break that down for the next 2 sprints? What tickets should each dev work on?” That’s the next step. All of these tools - the Slack bot, internal dashboards, HubSpot automation - were built using Cursor. I talk to Cursor or Claude, and they help me build. Mostly Cursor though. Before AI, I wouldn’t have been able to do any of this. I’d have had to rely on developers. Personally, I could have learned, but it would’ve taken 1–2 months just to learn to code enough. Now, I just talk to AI in English, and it builds things. I test to make sure it works. Cursor writes the code. If there’s an issue, I copy-paste the error, and it debugs with me. Same with optimization - it suggests improvements when I ask. The key is communicating clearly. If you don’t, the AI might interpret things wrong, and the output won’t match expectations. So planning, documenting, and clear instructions are critical. For hosting: the Slack bot runs on a cron job hosted on Render.com. Same for the internal dashboard and HubSpot integration (soon). For Jira/MCP, it’s just local on my machine - only I need access. The others are used by teammates, so they’re hosted externally. Some of these tools, like the Slack bot, took about a week to build. I built three bots for three different use cases - very quickly. With the dev team, that might’ve taken two weeks or more to spec, build, test, and deploy. So this approach saved a lot of time. Yes, AI can definitely replace low-level work. What I’m doing now might’ve required a junior dev a few years ago. Now, I’m that junior dev - maybe even mid-level on some days. That’s a game changer. I can build, ship, and test faster - and support my team better.

Render

dashboard

ChatGPT, Claude, and Cursor Turned This PM Into a One-Person Dev Machine

Too Long; Didn't Read

Get Free Unlimited IP Geo Data With IPinfo Lite

How One PM Replaced a Junior Dev with ChatGPT and Built Three Bots in a Week

About Author

Comments

TOPICS

THIS ARTICLE WAS FEATURED IN

Related Stories

ChatGPT, Claude, and Cursor Turned This PM Into a One-Person Dev Machine

How AI Has Impacted Product Management

Is AI Making Product Managers Obsolete?

Reimagining Product Management in the Age of AI

Is AI the End of Sustainability?

The Noonification: Use This 7-Step McKinsey Framework to Solve Any Problem (1/10/2023)

ChatGPT, Claude, and Cursor Turned This PM Into a One-Person Dev Machine

How AI Has Impacted Product Management

Is AI Making Product Managers Obsolete?

Reimagining Product Management in the Age of AI

Is AI the End of Sustainability?

The Noonification: Use This 7-Step McKinsey Framework to Solve Any Problem (1/10/2023)

Light-Mode

Classic

Newspaper

Dark-Mode

Neon Noir

Minty

HN StartUps