How two companies made agentic AI the core of modern enterprise workflows
Ryan Narragon · 16 November 2026
Read the originalLendi Group has AI agents turn project documents into user stories for coding agents, while CHG Healthcare spent two years on oversight before any agent reached production.
Ryan Narragon of Atlassian, the maker of the work-tracking product Jira, wrote up how two customers use AI agents. Lendi Group, an Australian home-loan company, moved the planning of 40 teams into Jira and uses agents for planning, coding and review. CHG Healthcare, a healthcare staffing firm in a regulated industry, built rules and oversight first and now runs more than 65 agents.
- At Lendi, an AI agent drafts a project document from recorded video walkthroughs, and other agents split it into user stories with acceptance criteria written for coding agents.
- Lendi says about 85 per cent of its software development is driven by AI agents, and that more than 16,000 manual actions were removed in six months.
- CHG Healthcare keeps its agent instructions as versioned files with dozens of revisions, so that any change can be rolled back.
- At CHG, a group of engineering, privacy and security leaders oversees the agents, and one agent scores the risk of proposed system changes for the change review board.
A product team that lets AI agents write user stories can keep those stories in the same tool as the rest of the work. There, people can review them before coding starts. That advice is an inference and is not the author's claim. The piece is a vendor's account of its own customers, and its figures are the companies' own.
Have an AI agent break an approved project document into user stories with acceptance criteria written for coding agents, and keep them in the team's work-tracking tool.
Keep AI agent instructions as version-controlled files so that any change can be reviewed and rolled back.
Derived by Working Surface from the article; more in the Playbook. Source line: We didn’t launch anything into production until we could prove the value.
17 November 2026: as AI agents and staff outside engineering produce the stories, specifications and tools that others build on, a team must name who checks them and who looks after them.