About
A Nordic Digital Marketing & Directory Services Company ran its AWS estate with cost governance resting on a single operations engineer. It rarely reached the top of the list, and waste drifted unseen.
Rather than present a proposal, Firemind ran a live deployment of its IT Operating Engine inside the client’s own AWS development and QA account over two months. Cost optimisation ran as a continuous part of operations, not a one-off audit.
Scope: the pilot ran on a single AWS development and QA account over two months, not the client’s wider production or corporate estate. All savings on this page relate to that environment.
Challenge
Cost governance was one of many tasks resting on a single operations engineer, and it rarely reached the top of the list. Without a recurring discipline, spend drifted:
- Idle resources kept running. Databases and compute provisioned for peaks billed in full while sitting near-idle for most of their lives.
- There was no feedback loop. Right-sizing happened once, if at all. Nothing watched the estate continuously to flag waste as it appeared.
- Savings claims needed proof. Before any commitment, the client wanted figures it could trust, verified against the real environment rather than estimated.
The ask was a continuous FinOps discipline surfacing real, verifiable savings, not an audit that aged on delivery.
Solution
Over two months, Firemind ran autonomous cloud operations on the client’s AWS development and QA estate, powered by its IT Operating Engine. Alongside patching, incidents and security, it ran cost optimisation continuously, reading the estate and turning underused capacity into a ranked, costed set of actions.
The live deployment proved four things:
-
It found the savings, item by item. The engine analysed the estate for waste and produced a ranked list of actions, not a vague percentage. It put a specific price on each one: idle databases, an over-provisioned search cluster, unattached storage and schedulable compute.
-
It captures the savings a human would skip. Manual FinOps only chases the big wins, because it is not worth an engineer’s hour to save ten pounds a month. The engine carries no such cost, so nothing is too small to include, down to an unassociated IP address worth about four pounds a month. Individually small, those items still make the list, and across a wider estate they add up.
-
It checked itself against the live estate. Firemind cross-verified the engine’s analysis against the real environment. The recommendations closely matched the actual state of the estate, so Firemind could state the savings as confirmed, not caveat them as estimates.
-
It corrected its own recommendations where the data demanded. Where the evidence pointed the other way, the engine adjusted. It downgraded one database from termination to a scale-down with an after-hours stop once the data showed 20% peak use, and when it flagged a test instance that Firemind itself owned, Firemind excluded it from action.
Every action executed inside the client’s own AWS account, audit-logged, with the client in control of what could auto-execute, what needed approval and what was blocked.
Results
Running live on a single AWS development and QA account over two months, the deployment produced a verified savings picture and a continuous discipline to maintain it. The largest confirmed savings, per month:
The six largest items shown. The engine also surfaced and priced the long tail of smaller compute, storage and networking savings, the items a manual review skips, with further headroom still to be assessed.
- Long-tail savings included. 714 GB of unattached storage, a second idle DocumentDB database, and smaller networking items that a manual review would not pursue
- A continuous discipline, not a one-off audit. Cost optimisation now runs as part of operations, catching waste as it appears rather than ageing on a shelf
- A funded path forward. The engagement has progressed into commercial business case discussions, with further savings headroom still to assess across the wider estate
The small recurring items are where autonomous FinOps compounds. Run continuously across a wider estate, it turns a one-off cost review into an ongoing feedback loop.
