What is CostOps?

How is CostOps different from FinOps?

FinOps is the cross-functional practice for managing cloud value through shared accountability between finance, engineering, and the business. CostOps is the execution layer that sits on top of it — agents that own the boring work so FinOps practitioners can focus on strategy and forecasts.

CostOps agents ingest billing, usage, tags, and architecture context, then propose diffs: Terraform changes, RI/SP coverage moves, autoscaler tweaks, idle cleanup, and Slack-based approvals. They track realized savings against forecast and learn from rejections, so every workload converges toward its efficient frontier without a human chasing each ticket.

What does a CostOps agent actually do day to day?

A CostOps agent runs an eight-phase loop every day across AWS and GCP: detect the anomaly, isolate the cost driver, trace the root cause, wash the data, run the chase, open the Merge Request with the fix, ship it under the approval gate, and learn from every approved change.

The unit of work is a landed change, not a dashboard recommendation. Each agent run produces an MR with the diff, the rationale, the projected savings, and the rollback. The team picks the approval gate per environment — notify, act-with-approval, or autonomous — and the agent keeps shipping inside that policy. Realized savings get attributed back to the original anomaly so the loop closes on data, not promises.

Why does CostOps matter now?

AI workloads broke the old FinOps loop. Token spend changes hourly, crosses model and vendor boundaries, and no longer maps cleanly to traditional cost categories. CostOps was coined to describe the agent-led closed loop required when humans can no longer reconcile usage fast enough.

The FinOps Foundation 2026 framework expanded scope to all technology value, including AI. CostOps is the agent-native execution pattern that complements that scope: it covers token usage as a first-class signal alongside compute and storage, attributes AI spend back to features, and ships fixes without waiting on a quarterly review.

FinOps vs CostOps vs AgenticOps vs Manual Cloud Cost

Four operating models for managing cloud spend. CostOps sits between human FinOps practice and the broader AgenticOps umbrella — specifically the closed-loop, agent-driven execution layer for cost.

How to adopt CostOps

CostOps deploys without ripping out FinOps. The CostOps Agent owns the toil; the FinOps team keeps strategy, allocation, and unit economics.

1. Step 1

   Instrument

   Connect billing, CUR/FOCUS, tags, and code repos so the agent has ground truth on what costs what, who owns it, and which service depends on it. CloudThinker Connections handles this in four network tiers.

2. Step 2

   Co-pilot

   Let the CostOps Agent draft Merge Requests and Slack approvals while a human keeps merge rights. Two-week burn-in per domain (idle cleanup, right-sizing, commitment laddering) before promoting any to autonomous.

3. Step 3

   Autopilot guardrails

   Graduate safe categories — idle cleanup, off-hours scaling, dev-account governance — to autonomous remediation under policy bounds. The audit log keeps the receipts. Senior reviewers focus on novel anomalies and forecast strategy.

Frequently asked questions

Who coined the term CostOps?

CloudThinker formalized the term with the CostOps Agent launch in 2026, framing it as the agentic execution layer above FinOps. The CostOps Agent runs an eight-phase daily loop across AWS and GCP: detect, isolate, trace, wash, chase, open MR, ship under approval, learn. The full launch post lives at https://cloudthinker.io/blogs/introducing-cloudthinker-costops-agent.

Does CostOps replace my FinOps team?

No. CostOps removes the toil — the rightsizing tickets, the idle-resource chase, the commitment ladder paperwork — so the FinOps team can focus on strategy, allocation, unit economics, and forecasting. The team grows in influence as the boring work goes away. CostOps and FinOps are complementary, not competitive.

Does CostOps cover AI and LLM spend?

Yes — token usage is a first-class signal alongside compute and storage. CostOps agents attribute AI spend back to features, services, and teams; surface anomalous prompts (looped agents, runaway batch jobs); and propose model-selection changes when a cheaper tier would meet the SLA. This is what separates CostOps from pre-2025 FinOps tooling, which generally did not model token spend.

How does CostOps integrate with Slack, Jira, and GitHub?

CostOps agents land their proposals where the team already works — Merge Requests in GitHub, GitLab, or Azure DevOps; tickets in Jira or Linear; approval threads in Slack or Microsoft Teams. Approval is per-environment, per-service. Realized savings get posted back to the same surface so the loop closes in public.

Is CostOps in the FinOps Foundation framework?

The FinOps Foundation 2026 framework expanded its scope to all technology value, including AI. CostOps is the agent-native execution pattern that complements that scope. It implements the framework's "automate" and "operate" capabilities through autonomous agents rather than human ceremony, and it covers token-level AI cost — a gap most pre-2026 FinOps tooling did not address.