AI‑agents deliver real it‑savings

It‑chefs still struggle to prove investment returns when they scale AI‑agents, yet some profitable use cases are emerging that can give cios concrete cost savings. Rhonda Baldwin, cio at Launch Darkly, says coding agents and service desk agents help the SaaS provider lower it‑costs. Coding agents increase technical capacity without a proportional rise in staff, and they also help optimise cloud costs and reduce workflow expenses, she says. In addition to coding and cloud‑cost optimisation, it‑chefs see level‑1 and level‑2 it‑support as another promising area where agents can free budget, even though many are just beginning to measure savings from AI‑agents and those who do use several metrics. For Baldwin, these savings include avoiding about one million dollars in two optimisation projects and an additional 120 000 dollars by building its own access‑management solution with AI‑coding tools. The company also saved roughly 50 000 dollars annually by using agent‑based AI for level‑1 it‑support. This not only boosts employee productivity but also avoids costs related to software, implementation, access management and hiring remote technicians, she says. It‑support and self‑service are strong use cases for it‑chefs who want to cut support‑agent expenses, says Gil Pekelman, ceo of AI‑agent vendor Atera, whose firm recently published a joint study with Forrester claiming a significant return on investment for its incident‑support agent. The calculation is simple. A company with 10 000 employees outsources all it‑support; if it shuts that operation now it can immediately measure savings, he says. Kevin Rooney, it‑chief at West Monroe Partners, reports substantial savings for the digital consultancy thanks to such an implementation. Employees can now contact an internal it‑and HR‑support agent in Teams or Slack to get answers to common questions, request software access, reset passwords, unlock accounts, submit and update support tickets or be escalated to a live support worker when needed, he says.

By handling large volumes of recurring requests and improving response and escalation, the it‑team can spend more time on complex problems and higher‑value work, he adds. West Monroe is now evaluating the bot’s impact on self‑service, ticket volume, response times and capacity savings, but it has already cut annual managed‑service‑provider costs by 40 percent, he says. The bot also saved an estimated 2 700 hours per year, he adds. Cloud‑cost savings Matthew Wallace, co‑founder and cto of Kamiwaza AI, has also seen large it‑savings thanks to agent‑based AI. The firm uses agents in software development, project management, data science and systems engineering. Cloud optimisation is another area where the firm saves money by using agents, he says. A set of agents monitors new cloud systems, coordinates their deployment and handles budget approvals, providing a simple way to stop unauthorised spending and log it. The agent‑based system replaced a function that required extensive human coordination and attention. You must check who configured it, when, what it costs, whether it has internal approval and details about resource sizing, placement and so on. Later you must verify who responded with information and whether any senior manager approved or rejected it, he says. The first batch of cloud‑optimisation agents that Wallace’s team implemented reduced cloud spend by 70 percent immediately. Some organisations overstate it‑savings from agent‑based AI because they do not measure all costs, says Jeet Pattanaik, founder and cto of Glokal AI, a provider of AI‑based solutions. He believes some it‑chefs do achieve real savings, but they are often smaller and take longer than promised, and most organisations are still chasing ROI. Pilot projects always look good because they handle simple cases, he says. Then odd cases surface in production, and someone must check what the agent actually did.