Atlassian

The missing layer between what your team decides to build and how your agents execute it. The bottleneck in AI-assisted development isn’t code generation. It’s intent. When you leave details out, the model fills in the blanks for you. Omit the scope, architectural constraints, and edge cases, and the agent quickly starts solving a completely different problem. When that happens, the agent isn’t just wrong – it’s confidently wrong, moving just as fast in a direction you didn’t intend.

Customer work rarely lives in one place. Salesforce holds the account, opportunity, contact, and case details. Jira shows the product work behind a customer commitment. Confluence holds the account plan, enablement materials, and meeting notes. Jira Service Management tracks support requests and escalations. Slack, Microsoft Teams, Google Drive, and other tools hold the conversations and assets that explain what’s really happening.

You know the drill when it comes to creating presentations. The afternoons spent creating a deck from scratch – aligning text boxes, copy and pasting images, moving the slides around to find the right flow, the list goes on and on. The problem isn’t that the story isn’t there. It’s that getting the story out onto slides takes way more time than it should. That’s where Confluence slides comes in.

Every organization is in the middle of an agentic transformation. Teams everywhere are rapidly building and deploying agents that access data, trigger workflows, and influence decisions, often without centralized oversight. That shift happened fast. Faster than most organizations planned for. Gartner ® predicts that “by 2028, an average global Fortune 500 enterprise will have over 150,000 agents in use.

Without the right context, every day is day one for a coding agent. Agents may have the intelligence needed to write, refactor, and review code, but they still face the same challenge developers do: understanding how complex systems actually work. They need the right context to navigate cross-team dependencies, ownership boundaries, and downstream impacts that are hard to understand from any single workspace. Code only tells part of that story.

The past couple of years have changed how my team operates. We’ve experimented, adjusted quickly, and seen meaningful early results. That momentum got me excited about where marketing is headed in the age of AI, even as we continue learning along the way. Since then, we’ve witnessed what holds companies back from succeeding with AI.

KFC’s engineering teams were drowning in release admin — chasing people for risk assessments, impact details, and component info. Not anymore. Now their custom Change Agent (built with Atlassian Rovo) pulls up everything teams need against a release — risk, impact, associated components — instantly. The result? Massively reduced admin time, engineers doing what they love, and a culture shift from “who do I ask?” to “I can find it myself.”

Every company investing in AI tools is asking the same question: how many people actually need to be good at this for it to make a difference? It’s a fair question. AI adoption is uneven. Some people use it constantly; others barely touch it. Leaders are left wondering whether the investment pays off unless everyone gets fluent, or whether a smaller critical mass is enough. My team, the Teamwork Lab, got a chance to test this at Atlassian.

@mention an agent on any page and it creates, edits, and comments alongside your team. Through the Atlassian Rovo MCP, the same agents work from Claude, Cursor, or your IDE. Agents have been working in Confluence since we launched custom agents in May 2024, and teams now run more than 5 million agent invocations a month. In February alone, Agents saved Atlassian customers more than 200,000 hours. Now, agents can do practically everything you can do.

We interviewed 15 high-throughput Atlassian engineers, identified through PR throughput data and peer nomination. All work on brownfield codebases: systems with years of history, broad surfaces, and layers of complexity. The most interesting finding: AI didn’t make their fundamentals obsolete – it accelerated them significantly. Those fundamentals are long-standing habits, but in an AI-accelerated world, the gap between engineers who apply them and those who don’t has become more noticeable.