Levelling up with AI: 8 Lessons from Atlassian's AI Product Builders Week

[#1]

Hello Atlassian Community 👋

I’m sharing my top 8 takeaways from Atlassian’s AI Product Builders Week, which is an internal event where teams across Atlassian came together to experiment, prototype, and share practical ways to use AI in our daily work, including how AI can be practical for anyone, how I use it day-to-day at Atlassian, what worked, what didn’t, and some simple ways you can get started.

TL;DR

• AI is transforming how we work at Atlassian, not just as a tool, but as a catalyst for faster learning, better alignment, and smarter product development.

• The biggest wins come from making AI prototyping a daily habit: start small, iterate quickly, and share learnings openly. This approach accelerates team alignment, surfaces real user needs, and helps us deliver value faster.

• To stay ahead, we need to keep experimenting, share what works (and what doesn’t), and always put our customers first.

• Making AI prototyping a habit is how we’ll build better products and faster.

---

1. The Quality Bar for Product Management Work is Increasing 🚀

AI tooling enables Product Managers to communicate product direction better and more often. Over time, this communication will be more necessary, because teams will be able to ship more software and delivering value faster to our customers.

2. Learn by Doing, Not by Perfecting 🏃‍♂️

One of the most valuable lessons came from watching teams dive into prototyping. The most progress happened when people stopped worrying about crafting the perfect prompt and just started experimenting. Rapid iteration beats overthinking. If your first attempt is not perfect, refine as you go. The key is to get moving and learn from each cycle.

If you’re looking for practical advice on how to write prompts, I shared my tips in this Loom video 🎥.

Another tip: avoid firing off too many prompts at once. If you start seeing hallucinations or off-target outputs, slow down and simplify. Reduce variables, clarify constraints, and iterate in smaller steps. You’ll get cleaner, more reliable results and waste less time debugging.

3. Make AI Prototyping a Habit, Not a One-Off, to Accelerate Alignment 🔄

One of the most powerful lessons was how quickly prototyping can bring teams together. Fast demos created tighter feedback loops with your teams and your customers. Using AI, I could move concepts from ideas to testable flows in a matter of hours. For the past year, I have made AI prototyping part of my daily routine. Whether it is sharing concepts via Loom, building quick demos, using AI to communicate complex ideas or speaking to customers, the habit of experimenting and sharing has paid off. This approach helped us spot issues early and align on solutions faster, no matter our role or discipline.

4. Community Multiplies Progress 🤝

During the course of AI Product Builders Week, we co-created a Capstone Project, which is a collaborative challenge where teams worked together to apply what we’d learned in real time. Mentoring during this Capstone Project was a highlight for me! I learned as much as I taught. Helping others surfaced new patterns and tools, and I met PMs and designers who approach similar problems in unique ways. The open and collaborative environment made it easier to learn from each other and accelerate progress. Generally, sharing learnings, whether through Loom videos, Slack updates, or quick demos, multiplies the impact of every learning.

5. Automate the Mundane, Focus on What Matters 🤖

Agents and automation help us handle repetitive tasks, like generating meeting recaps, sending thank-you emails and synthesising customer feedback. This is not just about saving time; it is about freeing yourself up to focus on the work that moves the needle. If you haven’t tried automating routine tasks with AI, start small and build from there.

6. What to Look for in AI Prototyping Tools ⚖️

When experimenting with different AI prototyping tools, I wanted to find out which ones could help me (and my team) move from idea to interactive prototype as quickly and effectively as possible. I focused on three main things: 1) how well each tool supported designing user journeys, 2) how closely the prototypes matched my vision for the final product, and 3) how easy the tools were to use day-to-day. Here’s what I found after regular use:

• Some tools excel at supporting the design of user journeys, allowing you to map out flows and interactions in detail, but may be slower due to more extensive testing or simulation features. If speed is a concern, look for options that let you pause or skip unnecessary steps once you’re satisfied with the experience.

• Others are optimised for rapid visual design, producing prototypes that closely resemble your final product. However, these may sometimes fall short in translating intended user behaviours into fully interactive elements - so while the visuals look great, not every feature may be functional out of the box.

• There are also tools that strike a balance between user journey design and visual fidelity, offering both intuitive interfaces and strong prototyping capabilities.

No matter which tool you choose, the key is to experiment and find what best fits your workflow and project needs.

7. Evaluating AI models 🧪

The LLM (Large Language Models) Evaluations session really highlighted how crucial it is to have a hands-on, structured approach to evaluating AI models. I learned that using clear prompts, templates, and real datasets makes it much easier to measure if an LLM is actually solving user problems. The session showed practical ways to generate synthetic data, cluster feedback, and even use LLMs to judge outputs, making the whole eval process faster and more objective. This approach helps us quickly spot what’s working, what needs tweaking, and gives us confidence to ship improvements without getting stuck in endless manual reviews.

8. Use AI Responsibly🛡️

Using AI responsibly is non-negotiable. It’s not just about what’s possible, it’s about putting extra care into and doing right by our customers. We have a responsibility to make sure we’re protecting our customer’s data, being transparent, and avoiding anything that could break our customers’ trust.

Final Thoughts 💭

AI isn’t just another tool. It’s changing how we work every day. You make progress by doing, not by waiting for everything to be perfect. Start small, share what you’re working on, and keep learning as you go. People who get the most out of this are the ones who aren’t afraid to try new things and focus on solving real problems for users.

You don’t need perfect plans or polished designs figured out before you start using AI. Try to use AI to streamline your workflow, generate new ideas, or solve problems in creative ways. The sooner you put AI to work on real questions, the sooner you’ll discover what actually helps.

If you want to increase your impact, whether you’re problem-solving, designing, building, or leading, consider making AI a regular part of your routine. And if you ever get stuck, just ask for help.