Atlassian: how AI turns product managers back into builders

The article, published by Atlassian in January 2026 with input from Aakash Gupta (Product Growth newsletter), argues that product managers who actually build AI tools develop better judgment than those who only observe from the sidelines. The central claim: fluency comes from doing, and the doing has a natural progression.

Atlassian organizes AI adoption for product managers into three stages. The first is prototypes — quick experiments that address a specific pain point. A PM notices that compiling weekly customer feedback takes three hours, so they wire up a simple summarization agent using a no-code tool. The goal is learning, not production. If the prototype proves useful, it moves to stage two: workflows. Here, validated experiments become repeatable processes. The PM chains the prototype into a proper flow with inputs, outputs, and enough reliability that teammates can use it too. The third stage is code — productionizing the workflow into a durable system, sometimes involving software development, sometimes just tighter integration between tools.

The article maps 15 specific AI use cases across the full product lifecycle. On the discovery side: monitoring customer feedback on social channels, synthesizing Zendesk tickets, analyzing Gong sales calls, extracting insights from community forums, and generating analytics signals. On the delivery side: creating interview guides, drafting PRDs, estimating feature scope, prototyping interfaces with V0 or Bolt, and building production features using RAG architecture. The list is notable for its specificity — it reads like a checklist rather than a conceptual argument.

Supporting data comes from Atlassian’s own product organization. Over 90% of its 450+ product managers use AI tools weekly. Nearly 70% use Rovo Chat and Agents regularly, with an average reported time saving of 40 minutes per day. The article uses this data as one reference point rather than the central argument, which prevents it from reading as a product pitch for Rovo.

The piece is useful for two audiences. For individual PMs, it provides a concrete growth ladder: start with a prototype that solves a real problem, build the habit of automation, and graduate toward code once the workflow has proven its value. For product leaders, it suggests a model for spreading AI fluency across a team — not through training workshops, but by creating conditions where building small things is safe and encouraged.

One caveat: the article leans on Atlassian’s own internal context, and the specific tools mentioned — Rovo, Jira Product Discovery, Confluence agents — are native to their stack. PMs working in different environments will need to transpose the framework rather than follow it directly.