Where AI Can Cut Operating Costs (And Where It Can't)
Where AI automation can lower operating costs, what the evidence shows, and how to estimate your own saving before you spend anything.
By Muhammad Shahbaz · · 4 min read
The short answer: AI can lower operating costs where work is repetitive, text heavy and high volume, such as customer replies, document entry and follow-ups. It is not a guaranteed percentage. IBM's 2025 CEO study found only 25% of AI initiatives delivered the return leaders expected. So estimate your own saving first, then start with one process.
Where AI automation usually saves money
Savings come from time, not from software magic. The tasks that tend to pay back first share three traits: people do them many times a week, the steps are mostly the same each time, and the input is messy text (emails, PDFs, chat messages).
- Customer support replies. The best public evidence is a study of 5,179 customer support agents. Agents with an AI assistant resolved about 14% more issues per hour on average, and about 34% more among novice and low skilled agents in the 2023 working paper (Brynjolfsson, Li and Raymond, NBER). That is AI helping people, not replacing them.
- Document and data entry. Reading an invoice, order or form and entering it into your system. This is repetitive, so it is a common first project.
- Lead follow-up and qualification. Replying fast, asking the same questions, booking a call.
- Internal reports. Pulling numbers from several tools into one summary.
What the evidence says about results
Gartner predicts that by 2029 agentic AI will autonomously resolve 80% of common customer service issues and cut operational costs by 30% (Gartner, March 2025). Treat this as a forecast, not a promise.
On the other side, IBM surveyed 2,000 CEOs in 33 countries. Only 25% of AI initiatives had delivered the expected return, and only 16% had scaled across the enterprise (IBM, May 2025). Gartner also expects more than 40% of agentic AI projects to be canceled by the end of 2027 because of cost, unclear value and weak risk controls (Gartner, June 2025).
The honest reading: AI saves money when the process is chosen well and measured. It wastes money when it is bought as a trend.
How to estimate your own saving
Do this on paper before you hire anyone.
- Pick one process.
- Count how many times it happens per month.
- Time one normal case, including checking and fixing mistakes.
- Multiply: cases per month x minutes per case x cost of one working minute.
- Subtract the monthly cost of the tool and any human review you will keep.
The result is a ceiling, not a guarantee. Automation rarely removes 100% of the work, because unclear cases still go to a person.
Where AI does not save money
- Processes that happen only a few times a month.
- Work that needs judgment, empathy or negotiation.
- A process nobody has written down. Automate a mess and you get a faster mess.
What to do next
Start with one process and a clear measure, such as minutes per case. Our AI Automation guide explains how to pick that first process. If you want a second opinion, the Stuck Media Automation Audit is a fixed $700 and is credited to a build if you go ahead. See pricing.
Before you commit, a fixed price audit can show which task to automate first. Our AI agents and automation service covers the build.
Sources
- Brynjolfsson, Li and Raymond, Generative AI at Work, NBER Working Paper 31161, later published in the Quarterly Journal of Economics (2025): https://www.nber.org/papers/w31161
- Gartner, Gartner Predicts Agentic AI Will Autonomously Resolve 80% of Common Customer Service Issues Without Human Intervention by 2029 (5 March 2025): https://www.gartner.com/en/newsroom/press-releases/2025-03-05-gartner-predicts-agentic-ai-will-autonomously-resolve-80-percent-of-common-customer-service-issues-without-human-intervention-by-20290
- Gartner, Gartner Predicts Over 40% of Agentic AI Projects Will Be Canceled by End of 2027 (25 June 2025): https://www.gartner.com/en/newsroom/press-releases/2025-06-25-gartner-predicts-over-40-percent-of-agentic-ai-projects-will-be-canceled-by-end-of-2027
- IBM Institute for Business Value, CEO study, press release (6 May 2025): https://newsroom.ibm.com/2025-05-06-ibm-study-ceos-double-down-on-ai-while-navigating-enterprise-hurdles
- cost-reduction
- ROI
- operational-efficiency
- AI
- case-studies
FAQ
Questions on this topic. Answered.
Can AI really cut operating costs by 60%?
No source we found supports a standard figure like that. Results depend on the process. Gartner forecasts a 30% operational cost reduction from agentic AI in customer service by 2029, and that is a prediction. IBM found only 25% of AI initiatives delivered expected ROI, so measure your own case first.
Which business process should I automate first?
Choose one that repeats many times a week, follows mostly the same steps, and has messy text input. Customer replies, invoice entry and lead follow-up are common starting points.
Will AI replace my staff?
In the strongest study we cite, AI assisted human agents and helped novices most. Plan for AI to take the repetitive part and for people to keep unclear and sensitive cases.
Comments
Questions or thoughts? Leave a comment.
Keep reading
More from the blog on similar topics
AI AutomationWhat an AI voice agent is, where it fits, its limits, hand-off to a human, US consent basics and how to start.
· 5 min read
AI AutomationA plain explanation of a fixed price automation audit: process mapping, tool inventory, ranked opportunities, a build estimate and a 7 day timeline.
· 3 min read
AI AutomationWhat an AI chatbot costs from Stuck Media in Multan: $700 audit, builds from $1,500. What changes the price and what to ask in a quote.
· 5 min read
Tell us what’s stuck. We’ll show you what to automate first.
A 30-minute call with the engineer who would build it. You leave with a clear next step, whether or not we work together. We reply within one working day.