Atlassian engineering blog

Loading
Walk into every customer meeting knowing what you promised
Atlassian describes how to connect Gmail and Google Calendar to its Teamwork Graph so that Rovo can compare customer commitments in email against Jira delivery dates, prepare meeting briefs, and draft follow-ups from Confluence notes. The post is a product walkthrough with example prompts and setup links, and it offers little technical detail on how the integration works.
Scaling enterprise security standards across the Marketplace
Atlassian's CISO outlines three expanded Marketplace security programs: a continuous crowdsourced bug bounty, a centrally managed annual penetration testing program run with Bugcrowd, and joint partner incident response with Forge telemetry. The post also introduces the Atlassian Enterprise Certified badge, which consolidates these signals for enterprise procurement teams.
How two companies made agentic AI the core of modern enterprise workflows
Atlassian describes how Lendi Group and CHG Healthcare moved from isolated AI experiments to agentic workflows connected to Jira, Confluence, and the Teamwork Graph. The post covers an agentic software development lifecycle, governance practices such as version-controlled prompts and staged production rollout, and measured adoption outcomes. It is a vendor-written customer story from a conference session, so its technical detail is limited.
Scaling Jira to ~6M Versions
Atlassian describes how Jira Search slowed down when customers accumulated nearly 6 million release versions, because resolving version functions like earliestUnreleasedVersion() loaded all versions into memory. The team reduced P99 resolution time from about 18 seconds to 92 milliseconds by resolving only the project-scoped versions, and cut unscoped SQL fetch times from over 30 seconds to under 700 milliseconds by chunking large ID lists to keep Postgres on its primary-key index.
3 lessons from rebuilding Atlassian’s product development lifecycle for people and agents
Atlassian's Chief People and AI Enablement Officer describes three lessons from redesigning the company's product development lifecycle around AI agents. The post covers agents preparing fixes for 52% of vulnerabilities, spec-driven development that shortened feature timelines, and a program where non-engineers shipped production changes with engineering review. It argues that teams should redesign handoffs between product, design, and engineering rather than simply adding AI to existing roles.
Synchronous vs Asynchronous Communication
This Loom-sponsored guide explains the difference between synchronous and asynchronous workplace communication and offers a five-question framework for choosing between them. It covers audience size, interaction level, urgency, complexity, and misinterpretation risk, and promotes video messaging as a bridge between the two. The content is general workplace advice with little engineering-specific depth and frequently points readers to the vendor's product.
In sales development, a human touch stands out
Atlassian's Teamwork Lab interviewed eight sales development representatives and others about how AI has changed their work. SDRs now hand AI the research and drafting prep, but every customer-facing message gets human review, and the time saved goes to calls, meetings, and rapport. The post is a sales-team workforce piece with little technical detail.
From change risk to root cause: AI-native operations for the AI SDLC
Atlassian announces AIOps features in Jira Service Management: Prevention Center for assessing change risk before release, and Incident Command Center for incident coordination and automated post-incident reviews. It also describes integrations with Datadog, Sentry, and Splunk. The post stays at the feature-overview level and gives little implementation or architectural detail.
Our Q1 FY27 letter to shareholders
Atlassian's Q1 FY27 shareholder letter from CEO Mike Cannon-Brookes reports subscription ARR growth of 22.5% y/y, cloud revenue growth of 30.5% y/y, and RPO growth of 44.4% y/y. It cites adoption metrics for Rovo agents, the Teamwork Graph, and the Atlassian MCP Server, and mentions the Agentic Multiplayer Protocol. The letter is primarily financial and strategic, with little technical implementation detail.
Supercharge the Teamwork Graph with People Context
Atlassian describes how people data from identity providers and HR systems such as Okta, Workday, and Microsoft Entra ID is connected to its Teamwork Graph to give Rovo and external AI assistants organizational context. The post is mainly a feature walkthrough covering people cards, org charts, and admin controls, with little implementation detail.
Meet Code Reviewer, now available with Jira
Atlassian's Code Reviewer, an AI agent now generally available with paid Jira and Teamwork Collection licenses, reviews pull requests in Bitbucket Cloud and GitHub, leaving inline comments with suggested fixes and checking changes against linked Jira acceptance criteria. The post explains how to enable automatic review and reports internal results, including a 31% drop in median PR cycle time and 39% of comments leading to code changes. It also covers a beta Standards feature and the isolated,...
More control over permissions in Bitbucket Pipelines: new features and upcoming changes
Atlassian introduces step-level permissions, short-lived container tokens (BITBUCKET_OAUTH_TOKEN), and git-only scopes for Bitbucket Pipelines to replace long-lived credentials and broad default permissions. The post also lays out dated breaking changes for branch restrictions, default step scopes, and the Run a pipeline API, with YAML examples for migrating existing pipelines.
Teaching a 4B Model to Think Like a Policy Expert
Atlassian fine-tuned NVIDIA Nemotron 3 Nano 4B in two stages, supervised fine-tuning followed by DPO, to act as a first-pass acceptable use policy filter ahead of a larger 20B reasoning model. The post reports precision, recall, and F1 against the baseline, plus latency and throughput gains from running the smaller model first, and notes that AI agents ran over 300 experiments.
Inside a context graph: how agents connect the dots across work
The post argues that agents need connected context, not just tool access or vector search, to make decisions like renewal discounts. It introduces the context graph, which maps who owns what, what was decided, and what currently applies, and it describes a bounded agent workflow that stops to escalate when authority is missing or conflicting. The piece is framed around Atlassian's Teamwork Graph.
Measuring and orchestrating AI across the full engineering lifecycle
Argues that engineering leaders should measure AI impact at the level of end-to-end system capacity rather than adoption metrics such as seat licenses or AI-authored code. It cites DX survey data showing PR size growth and lower change confidence, and proposes a four-layer framework covering value stream flow, quality guardrails, capacity allocation, and developer friction.
Agentic Pipelines is generally available: run agents in Cloud or on your own Linux runners
Atlassian announces general availability of Agentic Pipelines for paid Bitbucket Cloud plans, letting agents run on Atlassian cloud runners or self-hosted Linux Docker runners. The post covers a YAML configuration for choosing an agent provider such as Claude or Codex, four built-in system agents, and native Atlassian Rovo MCP access, with reported results such as a 95% first-run merge rate on vulnerability fixes.
For the third year running, Atlassian named a Leader in the Gartner® Magic Quadrant™ for Marketing Work Management
Atlassian announces it has been named a Leader in the 2026 Gartner Magic Quadrant for Marketing Work Management Platforms for the third consecutive year. The post frames the recognition around how its Teamwork Collection, Teamwork Graph, and Rovo AI features are positioned to connect marketing work, with a customer quote and general descriptions of internal marketing use. It offers no real technical detail on architecture, implementation, or measured results.
How enterprise addmins are taking control of cloud performance…..and why it matters
Atlassian introduces System Health, a set of admin tools for Atlassian Cloud that surfaces incident alerts and performance data. The post explains common causes of slow load times, including data growth, unused configurations, inefficient filters, and network proxies, and publishes performance bands and client-side benchmarks. Most of the content describes product features rather than implementation details.
Tackling agent sprawl across the enterprise
The post argues that agent sprawl is an identity and accountability risk rather than classic shadow IT, and that enterprises should scale AI agents under tiered autonomy, permission inheritance, and centralized visibility. It frames governance as a way to enable adoption, but much of the content is Atlassian product positioning and a customer story rather than transferable technical detail.