Why I Used my Agent to Create over 100 Tasks in Jira for a New Software Product Idea

If you’ve ever tried to understand something new, you probably know how learning really works: reading docs and watching tutorials only gets you so far. The real learning kicks in when you get your hands dirty with specific, concrete examples and actually play with the real thing.

That’s exactly why I fired up my agentic AI and had it create over 100 tasks in my personal Jira account for an entire (completely made-up) new software product that I came up with.

Going Big on Purpose

I didn’t just ask the AI for a couple of sample stories with a few tasks that I could enter manually into Jira. I connected my agent directly to my Jira account and told it to build out full epics, user stories, subtasks, sprints via the Scrum and Kanban method. Why go so big? Because seeing a small demo is neat, but watching an agent operate at scale inside a real platform shows you what it’s actually capable of.

By doing this, I got three wins in one:

1. A massive, tangible example of agentic AI working with external tools.
2. A crash course in how Jira actually works — its structure, issue types, workflows, boards, you name it.
3. A basic understanding of the main tasks a development team would need to complete in order to build a social networking platform with JavaScript React as the front end and Python and sql postgres for the backend and how they work together.

I chose OpenCode to accomplish this since it is one of the most popular, open-source AI coding agents designed to help developers write, edit, refactor, and debug code right from their computer or terminal. It acts as an autonomous pair programmer that understands your codebase and can iteratively execute multi-step tasks.

I chose Jira to build out all the tasks for my social networking platform concept since it's one of the most popular project management platforms out there, especially for development teams.

I chose JavaScript React as the front end layer, because it is uniquely suited for social networking websites because it was built explicitly for this purpose. Meta (formerly Facebook) developed React to manage the massive, real-time data updates and intense user interactivity required by their own platform.

I chose Python as it provides a fast, flexible service layer. It acts as a middleman between the app's interface and databases. Benefits include fast coding since it uses simple, readable text, 2. connects easily with machine learning, which sites use this for targeted user feeds, and 3. has a ton of prebuilt libraries such as frameworks like Django and FastAPI, which handle common site needs instantly.

I chose sql postgres as my database layer because it maps complex user connections (e.g., who follows whom) using a rigid, organized structure, guarantees that every like, post, and comment is saved perfectly, even during system crashes, and offers Full-Text Search to quickly find users and hashtags without needing separate search tools.

One of the unique things about my learning style is that I refuse to be limited by “but I don’t need this for work.”

Me and What Feels Like Magic

I’ve always been the type who loves to gain deeper insight, explore outside of the box, chase down weird concepts, and expand my knowledge — especially anything in the tech world. To me, technology is straight-up magic. Most people interact with it every single day (scrolling apps, using websites, ordering stuff online) without ever seeing what’s happening behind the curtain, the enormous complex world that makes it all happen.

For me it's difficult not to think, how is that webpage built? How is the software structured? What does the database actually look like? How do all these systems talk to each other?

Learning this feels like getting an all-access backstage pass. You step behind the scenes and suddenly the “magic” starts making sense. And the best part? The tech world is insanely deep and diverse. You can keep exploring and growing literally forever.

Agentic AI is the latest chapter in that adventure. It’s not just chatting with a smart model anymore — it’s about giving the AI real tools, real accounts, and real projects so it can do things.

From Tasks to Real Insight

Once the agent had populated my Jira instance with a full product backlog, I kept going. I had it:

• Generate reports analyzing all the tasks it created
• Identify potential gaps in the programming/implementation plan
• Give advice on sprint planning
• Suggest even more tasks and refinements

And I’m still playing/growing with it.

There’s something powerful about working with real data in a live system instead of abstract theory. It stops being “I read about agents” and becomes “I watched an agent plan an entire software project in Jira.” The knowledge sticks better. The concepts become muscle memory.

So… Why Did I Do It?

Because I love to learn. And with AI, I can learn, deeply, endlessly, and at a quicker pace than any university can accommodate.

I don’t need a work mandate or a perfect use case. I just need curiosity and a willingness to experiment. Or really I just need the tools and platforms that will give my curiosity an environment to thrive in. Connecting an agent to Jira and letting it build something real was my way of peeling back another layer of the tech curtain. It was fun, eye-opening, and genuinely useful.

If you’re on your own learning journey, I highly recommend doing the same: pick a tool you’re curious about, connect it to an agent, and go big. Don’t settle for tiny examples. Build the whole thing (even if it’s just a concrete concept). You’ll be amazed what you discover.