Overview
Environment
Azure + Power Platform
Engagement type
Funded Assessment
Duration
8 weeks
Scope
AI CoE framework design, governance, environment strategy, cost management, architecture review, support model, and knowledge transfer
Client engaged Sonata Software's AI center of excellence (CoE) team to assess, design, and operationalize a governed AI framework across their Azure and Power Platform ecosystem. Over eight weeks, the Sonata team conducted a comprehensive current-state assessment, defined a target operating model for the AI CoE, designed environment controls, and delivered an AI support model, equipping the client to scale AI responsibly and sustainably.
Customer snapshot
Industry
TMTE
Headquarters
Dublin, Ireland
Company size
4000 employees
Revenue
Approx $2B
The business challenge
- Fragmented AI initiatives across the organization, multiple teams running independent GenAI experiments (Azure-based pipelines, RAG implementations, agentic AI flows) without centralized oversight, resulting in duplication of effort and inconsistent quality.
- Recent 'bill shock' incidents caused by uncontrolled Power Platform and Azure OpenAI consumption, with no cost attribution mechanisms to trace spend to specific projects or business units.
- Security and compliance gaps in AI deployments – absence of environment segregation between experimental agent flows and production, exposing the organization to data security and regulatory risk.
- No formal AI governance structure – lack of defined roles, approval processes, policies, or a project intake mechanism to evaluate and prioritize new AI use cases.
- Absence of a structured AI support model – no defined SLAs, escalation paths, or integration with existing ITSM tooling (e.g., Jira / ServiceNow) for AI-related incidents and requests.
Impact of the problem
Cost
Unattributed Azure OpenAI and Power Platform cost overruns ('bill shock'); no visibility into AI spend by project or business unit
Cycle time and productivity
Delayed AI project approvals due to undefined intake process; engineering effort wasted on redundant initiatives
Risk exposure
Compliance and data security risk from ungoverned AI experimentation in production environments
PoC objective
- Design a comprehensive AI CoE operating model
- Establish governance, compliance, and cost control mechanisms
- Define scalable architecture and environment strategy
- Enable controlled experimentation and production deployment of AI solutions
- Create a sustainable AI support model for ongoing operations
The Sonata solution – what we did
Sonata delivered a structured, multi-phase approach to establish an AI CoE and governance framework:
- Conducted a current state assessment of existing AI use cases, architectures, and tools (Azure pipelines, RAG, agentic AI, Power Platform, Copilot usage)
- Designed a target AI CoE operating model, including governance structure, roles, and policies
- Defined environment strategy and provisioning approach across Azure and Power Platform with clear separation of experimental and production environments
- Developed cost management frameworks, including tagging strategies, dashboards, and monitoring approaches
- Delivered architecture recommendations to improve scalability, traceability, and performance
- Designed an AI support operating model, including processes, roles, and service structures
- Enabled knowledge transfer and handover to customer teams for operationalization
Microsoft technologies used
Azure OpenAI Service
Microsoft Fabric
Azure
Power BI
Power Platform
Azure Monitor
Microsoft + Sonata credibility
30+ years
of partnership
Microsoft AI Business Solutions Inner Circle member
Microsoft Frontier Firm Partner
How we did it
- Structured 8-week engagement with phased delivery (assessment → design → validation → handover)
- Weekly stakeholder workshops and reviews for alignment and feedback
- Combination of technical assessment, governance design, and operational planning
- Iterative validation of deliverables with client's leadership and IT teams
- Alignment with Microsoft best practices for AI governance, security, and architecture
Deliverables
- Current state assessment report
- Draft and final AI CoE framework document
- Environment and governance plan
- Governance policy guidelines
- AI support model and processes document
- AI project intake guide
- Cost management and reporting guide
- Transition plan
- Handover package (including all documentation and architecture diagrams)
- Knowledge transfer sessions
Results and benefits
| Metric | Improvement |
|---|---|
| AI governance maturity | Structured CoE framework with policies resulting in ~ 50% improvement in governance maturity |
| Cost management | Improved cost attribution dashboards and tagging resulting in ~ 10 to 15% savings in Azure/Power platform spend |
| Cycle time for approvals | Defined intake and approval workflows leading to reduced approval cycle time by up to ~ 40% |
Download: 
