Orange, the France-based telecom operator, has been tackling the hard part of FinOps: getting engineers to act on cloud spend. The company runs FinOps Clean Days, when engineering teams set aside delivery work for a day to clean up cloud costs together, with a leaderboard, prizes, experienced practitioners guiding newcomers, and visible results for sponsors. Combined with gamified hackathons, the activities helped the company’s 100-plus-person FinOps community earn an internal Net Promoter Score above 70. Orange FinOps lead Camille Marini describes the model in this article and explains how AI agents extend it beyond the core community.

FinOps is framed here as a business-change problem, not just a cost-reporting problem. Recent State of FinOps reports identify getting engineers to take action as one of the top challenges organizations face, and the article’s answer builds on two principles: Cloud FinOps is a shared responsibility, and the only path to that shared responsibility runs through communication and deliberate change. Orange created a FinOps Community of Practice with support from Google Cloud Consulting, ran it on standardized communication channels, and kept meetings actionable by sharing optimizations and billing updates. The Clean Days came from recognizing that agile teams rarely prioritize optimization against sprint backlogs; the company created protected, collaborative, and fun time instead. The practices map onto McKinsey’s four building blocks of change: conviction and understanding, formal mechanisms, role modeling, and talent and skills. Gamification and leaderboards make new behavior visible and give colleagues something to emulate; experienced practitioners mentoring novices builds skills inside the community; regular sessions build conviction about where money goes.

Community scale has limits. Orange‘s FinOps community has more than 100 people, but the organization has thousands of engineers, and no central FinOps team can reach everyone directly. AI agents extend the capability of a FinOps team in two ways: they take on complex, time-intensive activities that previously needed a human, and they reduce friction for individuals across the business. The suggested starting point is to find where engagement drops off in the FinOps lifecycle. An awareness gap calls for an insight agent that pushes real-time cost data into daily tools; a bandwidth gap calls for a remediation agent that surfaces quick wins as ready-to-merge code changes; a complexity gap calls for an orchestration agent that gathers data and simplifies reporting.

The article recommends a sequence: establish the community practice first, introduce read-only agents that inform and suggest, and only once those are established should teams build agents that execute changes. Direct action carries operational risk, so it needs careful management, and it is also where significant wins often sit. For tooling, teams that want to deploy quickly with minimal code can use the Gemini Enterprise App, a no-code environment for creating agents; developers needing granular control can use the Gemini Enterprise Agent Platform (formerly Vertex AI), which supports frameworks like the Agent Development Kit (ADK). Cloud FinOps, the article concludes, is moving beyond centralized reporting toward action that happens where the work does. Organizations getting there start with culture, then use agents to carry it further than one team could reach. Orange‘s numbers came from the community work, so building that foundation is the part worth copying first.