Product Managers' Roles Expand with AI Integration

Just a few years ago, professionals across various crafts expressed concerns that artificial intelligence would eliminate their roles.

The industry is now witnessing a significant trend characterized by the overlapping of traditional roles and the emergence of the 'AI builder' paradigm, with startups increasingly prioritizing 'builders' over specialized PMs or designers, a concept now normalizing across companies of all sizes.

The fundamental objectives of a product manager—identifying product-market fit, developing products that customers value, and ensuring the commercial viability of these products—remain constant.

Despite the rapid advancements in AI, PMs should prioritize genuine customer needs over hastily adopting every new AI trend observed on social media platforms.

Administrative tasks such as generating meeting notes, tracking action items, and providing status updates are increasingly managed by AI tools, freeing up PMs to focus on strategic endeavors.

Atlassian CEO Mike Cannon-Brookes leveraged the internal AI tool 'Rovo' to quickly ascertain project statuses, bypassing the need for traditional manual reporting from product managers.

A critical new responsibility for PMs is managing the organizational change associated with the widespread adoption of AI tools, enabling them to act as catalysts for acceleration by integrating model intelligence with the company's unique operational context.

Atlassian conducted numerous experiments, organizing teams in varied configurations and measuring diverse outputs to enhance speed, concluding that the optimal approach for a PM role is highly situational.

The specific function of a product manager is not static but rather contingent upon both the type of product and its current phase within the lifecycle, with the discussion focusing on three distinct scenarios: new features, zero-to-one products, and enhancements to large existing codebases.

Confluence successfully rolled out its 'Remix' and 'Slides' features by effectively utilizing AI tools, significantly streamlining the development process.

A product manager, aided by an engineering partner who developed a simplified coding harness, submitted 26 pull requests in a single month, surpassing most engineers on the team in terms of code contributions.

This PM focused on resolving user experience issues, enabling engineers to allocate their efforts to more complex, higher-level tasks and consequently reducing the iteration time for new features from six months to six weeks.

Product managers actively managed model evaluations, which resulted in a twofold increase in throughput, leveraging the Arise platform for efficient prompt debugging.

The company implemented automated design bug fixes by utilizing an AI coding agent that mapped Figma designs directly to the corresponding code.

Engineers experienced a significant reduction in test creation time, from half a day to just 10 minutes, thereby accelerating the development cycle.

The product manager's role evolved to include active coding, operating within an isolated repository to safeguard the integrity of the main codebase.

The success of this approach was attributed to an intentional contribution model and a willingness to question established processes, with the engineering manager ensuring that appropriate guardrails were in place for non-engineer contributions.

The 'Robo' project began as a zero-to-one initiative spearheaded by PM Josh and designer Kevin, who engaged in 'vibe coding' to quickly develop a functional alpha version.

This alpha was then used by internal customers, allowing for early feedback and iteration before additional engineers were brought into the project.

PM Josh initially contributed code to the project but recognized that this was not the activity with the highest leverage for accelerating the team.

By transitioning to a steering role, he could concentrate on setting direction, prioritizing tasks, and unblocking the team, which significantly expedited progress, and his technical background provided a deeper understanding of obstacles and improved communication with engineers.

Jira, with its two-decade-old codebase, faces complex requirements across data centers, cloud environments, and high-security standards like FedRAMP.

Product managers focused on integrating AI-first features to fundamentally reimagine the software development lifecycle within Jira.

The team achieved a threefold increase in throughput, successfully delivering 22 user-facing features in just 10 weeks by streamlining prototyping through automated coding agents triggered by Loom recordings transformed into work items.

AI-driven triaging further enhanced efficiency, as agents processed Slack feedback and over 900 customer insight videos to prioritize engineering fixes effectively.

This shift eliminates manual tasks such as compiling research, creating presentations from scratch, and performing manual updates, with the overarching goal of reducing synchronous meetings by leveraging AI tools that comprehend organizational context through Atlassian's 'teamwork graph'.

Ratings within the index range from 1 (curious) to 5 (pioneering), functioning as a professional development tool rather than a performance metric, with the expectation that PMs will achieve Level 3 across all categories, reserving Level 5 for specific strategic roles.

These programs feature insights from external guest speakers and foster internal peer-to-peer mentorship, requiring participants to complete a project—such as building agents or checking in code—to apply their newly acquired skills, a program that has involved over 1,000 employees and resulted in more than 120 new operational workflows.

Atlassian's metrics primarily focus on the number of pull requests (PRs) successfully deployed to production, rather than merely the number of PRs written, and measure outcome by tracking features from initial conception through delivery and customer usage.

The company emphasizes the importance of measuring throughput at both the individual team level and across the entire organization, continuously experimenting with various metrics each quarter to identify the most effective indicators of success and velocity.