Every FinOps engineer is worried that AI is going to steal their job. I’ve worried about it. But I’ve also experimented extensively with AI, and I’ve got a pretty clear sense of what it can and can’t do in a FinOps context.
The overarching theme of AI in a FinOps context is that it reduces the amount of time you spend on the most tedious parts of the job and increases the amount of time you can spend on the most strategic parts of the job.
Let’s dive into a few examples of each, and how you can use AI to augment your FinOps skills now.
3 Ways To Use AI For FinOps
1. Cleaning Up Messy Tags
The bedrock of FinOps is organizing your costs. Gotta know what you’re spending money on before you do anything else. Organizing (allocating) cloud costs can be extremely difficult, not least because it’s extremely tedious. Poor tagging leaves many resources unassociated with the right products, features, teams, etc., and going through cloud resources by hand is an excruciating exercise — even if it’s your full-time job, which it is for me, and which it probably isn’t for you.
How I use AI to help: Once I’ve got a customer’s full cloud spend at my disposal, I prompt Claude to look at it and propose an allocation framework that matches their business hierarchy. What this doesn’t do is get us to 100% coverage, right out of the gate. But what it does do is give me a directionally accurate framework I can propose to the customer as a starting point — and save me about five hours in the process.
You can do this too. Just give Claude your AWS cost and usage report (CUR) or equivalent for other cloud provider. Run this prompt:
Attached is my AWS Cost and Usage Report. I need a first-pass cost allocation framework I can pitch to this customer as a starting point.
– Look at the tags available at both the account level and the resource level. For each tag, report spend coverage (percent of total spend with a value) and flag consistency issues (casing, typos, sprawl).
– Recommend a 3 to 5 dimension hierarchy that mirrors how a business actually runs, something along the lines of Account then Environment then Business Unit / Team / Product. Note where account-level and resource-level tags overlap.
– Output a table with: Dimension, Source Tag, Coverage percent, Recommendation (Include / Optional / Skip), and a one-line rationale.
– Call out the biggest tagging gaps so I know where cleanup will have the highest ROI.
Directionally accurate is the goal. I will refine the framework with the customer.
2. Streamlining Optimizations
A lot of FinOps boils down to using the right tool for the job, not necessarily the most expensive one. The cloud is fluid, and when a feature moves from development to production, you often realize you can change the underlying infrastructure to save money without hurting performance.
But these optimizations are rarely simple. Differences in risk, potential savings, and human responsible for making the changes all slow down this process.
How I use AI to help: Once you’ve got a list of potential optimizations, you can have AI sort them by these factors:
- Potential savings
- Difficulty/risk
- Human responsible
The third factor depends on having good cost allocation, which step 1 should have helped you start, and which CloudZero specializes in perfecting.
For low-effort, low-risk optimizations, you can even have AI do the work for you. Large language models (LLMs) are experts at using command-line interfaces (CLIs). Once you’ve had a human validate the savings opportunity, that human can authorize an LLM to do the work and realize the value.
Attached is my list of cloud cost optimization opportunities.
For each one:
- Estimate annualized savings in dollars.
- Rate difficulty and risk on a 1 to 5 scale. Name the specific risk factors (downtime, data migration, irreversibility, architectural change).
- Assign the human responsible by looking up the resource owner in the tags or allocation dimensions. Flag any ambiguous ownership.
Output a prioritized table with highest-value, lowest-risk items at the top.
3. Customer Profitability
The peak of FinOps is making strategic assessments that help you maximize the return on your cloud investment. This is hard to do if you’ve got messy costs or don’t know which engineer is responsible for what. But once you do, you can subject your cost footprint to complex questions in simple language that get you rich, unit-economics-level insights.
Questions like, “How profitable is Customer A versus Customer B?” An AI-powered solution can compare cost data and revenue data for the two customers and provide a simple table that shows customer profitability.
How AI can help: The CloudZero AI Hub has an MCP server that lets you analyze cost data alongside any and all other data — revenue data, customer usage data, observability data, engineering data, etc. — and a Claude Code Plugin that lets you do all the analysis within Claude Code, where you’re already working.
Other questions you could ask:
- “Why did costs spike last week?”
- “What’s driving Anthropic spend?”
- “Which customers are costing us the most to serve? The least?”
- “Which team owns this Kubernetes cluster?”
Research Report
FinOps In The AI Era: A Critical Recalibration
What 475 executives told us about AI and cloud efficiency.
Waste Less Time On FinOps Tedium
What anyone can do with AI today is eliminate a lot of the tedium at the Walk stage of FinOps.
- Generate a directionally accurate allocation framework
- Triage (and execute the simpler) cost optimizations
What you can do with AI once you’re a CloudZero customer is jump into the AI Hub and ask strategic questions that give you actionable business insights. CloudZero gives you 100% cost allocation, direct engagement with every engineer, and simple integration with the rest of the data about how your business is operating.
FinOps In The AI Era: A Critical Recalibration
What 475 executives told us about AI and cloud efficiency.
