How Lendi Group and CHG Healthcare moved beyond isolated chatbot experiments to build automated, connected systems of work that scale.
The learnings in this blog are based on the session, “How customers operationalize agentic AI with Rovo”, presented at Atlassian’s Team ’26 conference. You can check out this session and others on demand.
Top 5 takeaways for scaling agentic AI
- Design for enterprise-wide ROI: Prioritize measurable outcomes beyond individual productivity, as Lendi Group did by eliminating more than 16,000 manual actions in six months.
- Start small and build with teams: Prototype one focused workflow with the people who use it, validate its value, and expand responsibly.
- Make adoption a leadership priority: Leaders must guide changes in how teams work, not simply oversee technology deployment.
- Connect agents to shared context: Ground AI in current documentation, decisions, and work items to reduce manual handoffs across planning and delivery.
- Build governance into scaling: Follow CHG Healthcare’s approach of establishing security and privacy guardrails, proving value before production, and version-controlling prompts for repeatable deployments and safe rollbacks.
Many organizations start their AI journey with standalone initiatives such as drafting emails, summarizing notes, or testing ad hoc chatbots. While these tools offer individual productivity boosts, they often leave AI disconnected from where real work happens.
Across industries, forward-thinking enterprises are shifting from an AI-adjacent posture to an AI-native operational model by weaving agentic AI directly into their core business workflows.
During a recent discussion exploring how enterprises operationalize agentic AI with Atlassian Rovo, leaders from Lendi Group and CHG Healthcare shared how they transitioned from isolated experiments to scalable, automated workflows. Despite operating in vastly different sectors, fintech and healthcare staffing, both organizations followed a strikingly similar playbook to optimize AI across their operations.
From side projects to a connected system of work
A connected system of work brings planning and execution together to maximize context and outputs you can trust. The Teamwork Graph connects conversations, documentation, tickets, and goals so agents can reason across the work lifecycle.
Lendi Group: Powering an AI native software development lifecycle
Lendi Group, one of Australia’s fastest-growing fintech platforms managing over $100 billion in home loans, established an executive mandate to become fully AI-native. For Lendi Group, that meant embedding agents into everyday decision-making and execution paths while empowering every employee to act as a leader and coach for their digital agents.
Previously, its teams planned work across six separate tools, and preparing their quarterly plan took three days of manual effort. Now Lendi Group uses Atlassian Goals and Jira Plans to help 40 cross-functional teams track progress continuously, with teams updating their work on their own schedules rather than waiting for the next quarterly planning cycle. The guiding rule is simple:
“It needs to be in Jira.”
In this mature environment, roughly 90% of agent interactions are triggered automatically by system events and workflow transitions, without manual prompts. Agentic workflows eliminated more than 16,000 manual actions in six months, creating measurable operational capacity.
That is incredibly powerful because that is work that is being done without a manual trigger. It is happening while people are away, not at their desk.”
Matt Hargraves, Head of Product Delivery and Automation at Lendi Group
At the time of the session, Lendi Group reported that 85% of its software development was driven agentically, combining Rovo with coding and prototyping tools such as Cursor, Claude Code, and Magic Patterns. The organization was beginning its first full end-to-end agentic SDLC run, bringing together the following workflows:
- Planning and requirements: Planning conversations captured in Loom video transcripts feed a Rovo agent that produces a structured project initiation document in Confluence. Subsequent agents break that document down into Jira epics and refine them into user stories with clear acceptance criteria tailored for development agents. As Matt Hargraves explained, “We’ve changed the way we look at acceptance criteria because it’s not a human doing the work. It’s an agent that’s picking up the work, and the agent that’s refining that work is doing it for another agent as well.”
- Code generation and deployment: Development agents turn structured backlog items into initial code, while connected pipelines track deployments and draft release communications.
- Marketing and legal compliance: In non-engineering workflows, marketing submissions through Jira Service Management trigger review agents that check assets against branding and legal requirements, helping reduce manual review work while keeping humans involved at every stage.
- People operations: Employee parental leave requests submitted through Jira Service Management are analyzed by specialized agents who calculate leave entitlements within minutes rather than days, enabling human teams to review decisions quickly and support employees during vital life transitions.
Empower non-technical teams to prototype their own solutions. Bring them into the fold. We uncovered some incredible insights and ideas that engineering teams and the teams working with AI every day just did not think about. They were incredibly valuable for us.”
Matt Hargraves, Head of Product Delivery and Automation at Lendi Group
To keep organizational context pristine, Lendi Group cultivates deliberate behavioral habits. Leaders summarize verbal decisions aloud at the conclusion of recorded meetings and post recaps of ad hoc conversations in shared Slack channels so the Teamwork Graph captures every nuance. As Matt Hargraves observed when reflecting on daily communication habits:
AI has changed how people work. An example of that is I find myself now in meetings, recapping deliberately, because I know it needs to be picked up by the transcript. Being really, really clear and deliberate in that action is done. Similar to things in Slack, I will actually recap and verify the ask. If someone has rang me ad hoc and updated me and changed compared to what was in Slack, then I make sure that is reflected in Slack itself, so then my Teamwork Graph knows that information has changed.”
Matt Hargraves, Head of Product Delivery and Automation at Lendi Group
Additionally, automated reminders prompt document owners to review content for accuracy every six months. Inaccurate content is archived, while early-stage thinking is made private to keep shared knowledge relevant.
Lendi Group also built dashboards to track AI adoption, agent volume, and return on investment for executive leadership and investors. That visibility makes it easier to assess whether agentic workflows are delivering value beyond individual productivity gains.

CHG Healthcare: Governance, trust, and operational excellence
As a leading healthcare staffing organization operating in a heavily regulated space, CHG Healthcare approached agentic AI with deliberate governance and a focus on enterprise reliability. It spent two years building technical guardrails and governance processes before deploying agents in production.
This structured approach fueled rapid growth, scaling from zero to over 65 active agents, logging more than 22,000 interactions, and achieving 135% adoption growth across the company.
We didn’t launch anything into production until we could prove the value. It was not vaporware. It’s not, ‘Let’s cross our fingers and hope it works and hope we beat our competitors.’”
Brad Spackman, Director of Enterprise Application Services at CHG Healthcare
To illustrate the cultural mindset required for this shift, CHG Healthcare cites the soccer-goalie analogy from Fred Kofman’s book Conscious Business.
Brad Spackman, Director of Enterprise Application Services at CHG Healthcare, puts the distinction between tasks and outcomes this way: “A goalie’s task is to protect the net. The job is to help win the game. As an engineer, is your job to write code or to write code to solve complex problems? As a BI analyst, is your job to write reports because it’s fun and you like data, or is it to improve decision speed and quality to help move your business forward?”
Teams should consider prompt precision and token usage alongside output quality to avoid unnecessary AI processing.
The production agents below show what that practical mindset looks like:
- Change quality evaluator: The agent evaluates proposed changes against historical knowledge in Confluence and generates a risk score to support more informed change advisory board discussions.
- Executive incident reporting: During major incidents, agents help turn incident information into executive summaries organized around condition, action, and need.
- Incident predictor agent: At the time of the session, CHG Healthcare was developing an agent intended to help predict service disruptions before they affect end users.
- Fractional project manager: To support technologists without formal project management training, the agent helps create project charters, RAID logs, and change-management logs, then generates structured Jira work items.
To ensure security and enterprise trust, CHG Healthcare established an AI technical center of excellence composed of engineering, privacy, and security leaders. They pair robust guardrails with continuous enablement, hosting open office hours and running minimal viable teams tasked with exploring innovative solutions exclusively through AI tooling. Furthermore, engineering teams manage agent prompts as version-controlled markdown files across dozens of revisions, ensuring safe rollbacks and repeatable deployments.
You want to make sure that you’re doing things with people, not to them. Partner with people. Think through the problems. What is my goal? What is the goal of the goal? What do we want to accomplish, and how can it actually be measurable?”
Brad Spackman, Director of Enterprise Application Services at CHG Healthcare

Strategic guidance for immediate execution
For organizations looking to scale agentic AI across their teams, both leaders suggest focusing on enterprise-wide return on investment rather than getting stuck optimizing hyper-specific, isolated workflows.
Regardless of your organization’s scale, there are critical reflection questions you can ask now to take action and achieve results similar to those of Lendi Group and CHG Healthcare.
- What is your immediate first step? Choose one workflow with the people who use it, then test and refine it together.
- How should you manage knowledge hygiene? Implement automated review reminders and clear archiving policies to ensure your documentation reflects the current operational truth and supports reliable agent decisions.
- How do you engage non-technical stakeholders? Invite them to test early prototypes and use their feedback to identify needs the engineering team might miss.
- How do you address organizational anxiety around change? Frame AI as a supportive coworker that eliminates tedious administrative chores, enabling employees to focus their energy on creative strategy and high-value collaboration.
Put AI to work with Teamwork Collection
Agentic AI delivers more value when it can draw on the goals, knowledge, and work your teams already share. Teamwork Collection brings Jira, Confluence, Loom, and Rovo together so teams can connect context, streamline handoffs, and build AI-powered workflows around real work.
Ready to move beyond isolated AI experiments? Explore Teamwork Collection and start with one workflow your team can improve together.

