IDC’s research VP for software development described low-code and no-code platforms this summer as the primary way companies build and run AI agents at scale. Microsoft Power Platform has 56 million monthly active users, up 27 percent in a year. That platform-wide count covers more than confirmed AI agent builders.
Gartner projects that by 2029 more than 80 percent of new business applications will be built on AI-powered low-code and no-code platforms, compared with 20 percent in 2024. These tools give operations teams a way to participate in building the systems they use, with technical support appropriate to the data and actions involved.
What the company examples report
Navien, a global HVAC manufacturer, reported saving 28,000 hours annually through AI agents and automation. Thirty-two percent of employees use self-service analytics to make decisions without calling IT. The account describes people close to the work building solutions with the tools available to them.
At Grupo Bimbo, two AI agents were built during an internal hackathon using Microsoft Copilot Studio. The reported results include a 20 percent drop in planning-phase audit time and risk and control matrix creation moving from days to seconds. Those are specific audit tasks, rather than a measure of the whole audit process.
EY’s reported figures cover a broader program: Microsoft 365 Copilot deployed to 150,000 employees, a 15 percent productivity gain, and expansion across more than 400,000 people globally. The account also cites 95 percent faster lead times, 37 percent lower finance operating costs, and up to 90 percent less manual work across core processes. These measures need their individual scopes checked before they can be attributed to a particular tool or builder group.
Atos reports operating 19,000 AI agents through a unified model built on Copilot Studio and Power Platform. Its figures include a 68 percent reduction in incident triage time, laptop compromise investigations moving from 25 minutes to eight, and 337 hours saved per week on investigations. Security operations staff did not shrink; twenty percent moved from incident response to governance, risk, and compliance work.
Give the process owner a bounded build
Choose a repetitive process and involve the person who knows its exceptions. Give that person access to an approved low-code builder, agree on a one-week test, and have technical staff review the permissions and connections before the system acts on live data. The test should reveal what the tool can handle and what still requires engineering or human review.
Drag-and-drop orchestration can reduce custom development. Built-in governance features still need configuration and testing against the company’s requirements. The builder’s knowledge of the job helps define those tests; it does not remove the security team’s responsibility.
Gartner projects worldwide spending on low-code development technologies at $58.2 billion by 2029. For a team deciding what to build now, the useful evidence is smaller: a working task, its failure cases, and the effort needed to run it safely.