What is a Cloud Center of Excellence?
A Cloud Center of Excellence (CCoE) is a cross-functional team that sets cloud standards, governs cost and security, and guides migration and modernization decisions for the rest of the company. In large enterprises it typically includes FinOps, cloud security, site reliability and architecture specialists, who turn the company's business goals into the architecture, cost and security decisions that deliver them.
Why don't most companies have one?
A Cloud Center of Excellence is expensive to staff. A FinOps engineer, an SRE and a cloud security engineer cost about $500–600K a year, fully loaded (Sherpa estimate), before an architect or a migration lead. Most growing companies on AWS run their cloud with a few engineers who also build the product, so cost, security and operations get reviewed when something breaks rather than continuously.
The gap shows up in the numbers. One in three IT leaders say more than half of their cloud spend is wasted (VMware Private Cloud Outlook, 2025). 99% of cloud security failures come from the customer's own settings (Gartner). 69% of companies run six or more tools just to watch their systems (Omdia, 2026).
How is an AI Cloud Center of Excellence different?
An AI Cloud Center of Excellence keeps the job of a CCoE and changes who does the routine work. AI agents read the company's own cloud data continuously and hand the team answers and fix plans; people keep the decisions.
| Traditional Cloud Center of Excellence | AI Cloud Center of Excellence (Sherpa) | |
|---|---|---|
| Who does the work | A team of FinOps, security, reliability and architecture specialists | AI agents working across your data, with your engineers deciding |
| Business context | In people's heads, documents and meetings | Recorded in one platform, with every decision and its history |
| Coverage | Periodic reviews of selected accounts | Every connected AWS account: cost rebuilt nightly, inventory scanned daily, security on the 1st and 14th of each month |
| Output | Reports, tickets and meetings | Answers in plain English and a fix plan for every finding |
| Cost | About $500–600K a year in salaries for three specialists (Sherpa estimate) | An annual subscription |
| Who applies changes | The team | Your team approves and applies every change |
What does an AI Cloud Center of Excellence need?
An AI Cloud Center of Excellence needs three things working together, which is why dashboards alone are no longer enough:
- Context: the business behind the cloud, including accounts and environments, owners and tags, and the history of decisions already made.
- Telemetry: cost, security, inventory and performance data from every account, in one place. In Sherpa, cost comes from AWS Cost and Usage Data and performance from CloudWatch and OpenTelemetry.
- Multi-domain agents: specialist AI agents that answer across cost, security, inventory and observability, and write the fix plan.
In Sherpa, the three meet in the Sherpa Knowledge Graph, which lets a team ask one question across every domain and get one answer grounded in its own AWS data.
What should an AI Cloud Center of Excellence not do?
An AI Cloud Center of Excellence should not change a company's cloud on its own. Sherpa proposes every change and the company's team approves and applies it; each decision is recorded with its history. It should also be honest about what it measures: Sherpa shows readiness against CIS 3.0 and SOC 2 controls, which is not an audit or a certification.
How does Sherpa work as an AI Cloud Center of Excellence?
- Ask: one chat across cost, security, inventory and performance, through the Sherpa Knowledge Graph.
- See: Insights dashboards from the executive summary to the single resource.
- Fix: Automate writes a prioritized fix plan for every finding.
- Decide: Deep Research answers strategic questions with a findings-first report.
- Operate: Sherpa Observability shows metrics, logs and traces next to cost.
- Move: the Migration Assessment turns a server inventory into a costed plan (paid add-on).
Companies that manage cloud costs well cut them by as much as 20–30% (McKinsey, 2023). A published Sherpa case study shows the same approach in a migration: 429 servers reviewed, 350 in scope, $630,987 over three years, every figure traced to a source.
