AI FinOps
Being an undifferentiated, recurring task, AI agents are coming to cloud financial operations. Dealing with financial data, and custom systems, the devil will be in the detail.
Problem
With AI, we need to be very careful that we are not adding to the cloud waste problem, rather than solving it. Many cloud users and suppliers alike are looking at ways to implement AI workflows into their cloud financial operations. The outcomes we all strive for with AI in FinOps are mainly to inspect, accelerate and action. Especially the agentic owned actions, today is still a sea of hopes and illusions, whatever vendors may claim.
Old way
Whilst large language models (LLM) are good (and getting better) at quickly processing and analysing vast amounts of data, one of the biggest challenges with just slapping a LLM on top of your existing data, is that it is going to yield the same generalized recommendations as a good FinOps person yields, including hallucinations. This is because the LLM is only as good as the data or query it receives. It is also not trained to tell you when it does not know something. Fast forward, and again you have analysts having to interpret, extrapolate and quality check the data from LLM's.
In most cases the goal is to automate recurring tasks. This could be anything ranging from running a commercial optimization analysis/renewal, to anomaly detection, to enforcing technical best practices for cloud deployments. The next step is to assess whether you want to have the AI agent own the action, or merely inform you. Most tools today are just capable of informing you. With questionable input data and the action still being on the human/user to implement it, the results are usually the same as in the traditional FinOps world as well: inaction.
"The problem and the solution"
New way
We believe, if implemented correctly, AI can be a great way for organisations to refocus on their core competencies and outsource some undifferentiating recurring tasks to AI agents. We also think today it can already increase accessability of financial data to the wider organization, beyond FinOps and engineering, to understand what is going on with their enormous cloud bill and cloud waste.
All of Optimaze's AI functionality are natively part of the product and are outcome focused. Optimaze's develops its agentic workflows with the following tenets:
AI with purpose - AI should be a natural part of the solution, not a bolt on. At the same time, sometimes you do not need AI, a simple automation would do the trick.
Fine grained control - without control, AI can lead to unwanted outcomes, including more bill shock. With Optimaze you will always have detailed control over key parameters.
Confident - Especially in dealing with finances, you need to be able to trust what the AI is telling you or doing, to the comma.
Secure - Even if your cloud financial data does not have high security classifications, Optimaze treats this data as top secret, by segregating, masking and encrypting as standard.
For more information on our development direction, and a preview of our features in private beta, we'd love to talk to you.
Atlas bridges the gap between engineering and finance teams. It delivers clear, actionable insights based on enhanced attribution data, in a language that makes sense to whoever is asking.