AI Agents Deliver Measurable IT Savings at LaunchDarkly

LaunchDarkly CIO Rhonda Baldwin says coding agents increase engineering capacity without a proportional headcount increase while also helping optimize cloud costs and reduce workflow consumption costs. [1] She says the company used coding agents to build internal tools instead of purchasing more software, avoiding about $1 million across two optimization projects and another $120,000 through an internally built enterprise asset-management solution. [1]

LaunchDarkly also reports about $50,000 in annualized savings from agentic AI for tier 1 IT support, alongside avoided costs tied to software, implementation, asset management, and remote service technicians. [1] CIO.com says tier 1 and 2 support is viewed by IT leaders as a promising area for freeing budget, although many organizations have only begun tracking agent savings and use different metrics. [1] Atera CEO Gil Pekelman describes IT support savings as straightforward to calculate, citing a 10,000-employee company with offshored support that could measure savings immediately if it closed that operation. [1]

The evidence is narrow but concrete: LaunchDarkly is counting avoided software purchases and annualized support savings rather than relying on one abstract ROI figure. Atera offers a different frame, treating the removal of outsourced support as an immediately measurable saving, while LaunchDarkly’s examples span coding, internal software creation, and tier 1 service. For CIOs, the practical lesson is to separate cost avoidance from recurring savings and document each use case with its own metric. What the sources don't address: how deployment and ongoing operating costs affect net returns.