Introduction

TODO: 2-3 paragraph introduction explaining what the AI Readiness Framework is and who it's for. Cover the problem it solves, when to use it, and what outcome it produces. Write for a VP/C-suite audience who needs to evaluate whether their organization is ready for AI investment.

TODO: Second paragraph covering the framework's origin — developed from Xephyr's work across 40+ AI strategy engagements, distilling common blockers into a repeatable diagnostic tool. Explain that the framework produces a scored readiness profile across five dimensions that guides prioritization decisions.

TODO: Third paragraph on how to use this framework — as a self-assessment tool, in a facilitated workshop, or as a pre-engagement diagnostic before a formal AI strategy engagement with Xephyr.

Phase 1: Data Foundation Assessment

TODO: Description of Phase 1 — evaluating data collection, storage, quality, and accessibility. This is typically the most critical gate: organizations without reliable, accessible data cannot build reliable AI systems.

Audit Data Sources

TODO: Describe how to inventory all data sources across the organization, evaluate quality, accessibility, and governance status. Include guidance on scoring each source on a 1-5 scale for completeness, accuracy, timeliness, and consistency. Explain what a "healthy" data source profile looks like versus one that requires remediation before AI work can begin.

Assess Data Infrastructure

TODO: Describe how to evaluate current data infrastructure — cloud vs. on-prem, data lake vs. warehouse, real-time vs. batch capabilities. Include a decision matrix for determining whether infrastructure is AI-ready or needs modernization. Cover key questions: Can you run ML training jobs? Can you serve model predictions at scale? Is there a feature store?

Evaluate Data Governance Maturity

TODO: Cover data governance assessment — data ownership, access controls, lineage tracking, quality monitoring. Explain the link between governance maturity and AI risk management. Include a checklist of governance capabilities required before deploying AI in regulated industries.

Phase 2: Talent and Capability Assessment

TODO: Description of Phase 2 — evaluating whether the organization has the human capital to build, deploy, and maintain AI systems. Cover the full spectrum from data scientists to ML engineers to AI product managers.

TODO: Data science team structure and skill gaps identified against project requirements
TODO: ML engineering capabilities mapped — can the team deploy, monitor, and retrain models in production?
TODO: Data engineering capacity assessed — do pipelines exist to feed models with clean, timely data?
TODO: Executive sponsor identified, committed, and able to remove organizational blockers
TODO: Change management plan drafted for teams whose workflows will change post-AI deployment
TODO: Training and upskilling roadmap created for capability gaps that can be filled internally

Phase 3: Infrastructure and Tooling Readiness

TODO: Description of Phase 3 — evaluating the technical infrastructure and tooling stack for AI/ML workloads. This covers compute, storage, orchestration, and monitoring.

Assess Compute Resources

TODO: Evaluate compute availability for model training and inference. Cover cloud vs. on-prem trade-offs, GPU requirements for different model types, cost modeling for training vs. inference at scale. Include guidance on when managed ML platforms (SageMaker, Vertex AI, Azure ML) are the right choice.

Review MLOps Tooling

TODO: Assess the MLOps tooling stack — experiment tracking, model registry, CI/CD for models, monitoring. Explain what a minimal viable MLOps stack looks like for organizations just starting out versus what a mature stack looks like for organizations running many models in production.

Phase 4: Governance and Risk Framework

TODO: Description of Phase 4 — evaluating the organization's ability to govern AI systems responsibly, manage model risk, and comply with relevant regulations.

TODO: AI ethics policy drafted and reviewed by legal, compliance, and executive team
TODO: Model risk management framework defined — who approves models before production?
TODO: Bias and fairness testing protocols established for relevant use cases
TODO: Explainability requirements identified for regulated or high-stakes decisions
TODO: Incident response plan created for model failures, data breaches, or adverse outcomes
TODO: Regulatory compliance requirements mapped — GDPR, CCPA, sector-specific AI regulations

Maturity Levels

Level 1Ad Hoc

TODO: Organization has no formal data strategy. Data is siloed, undocumented, and accessed inconsistently. AI experiments are isolated and rarely reach production. No dedicated data or AI team.

Level 2Developing

TODO: Basic data governance in place. Some standardization of data collection and storage. Early analytics capabilities with dashboards and reporting. First AI experiments underway, but deployment path unclear.

Level 3Defined

TODO: Formal data strategy aligned with business goals. Centralized data platform with documented pipelines. Regular ML model deployment in at least one business domain. Cross-functional data team with defined roles.

Level 4Managed

TODO: Advanced analytics in production across multiple domains. ML models deployed with monitoring and retraining pipelines. Data quality monitored automatically. AI governance framework established and followed.

Level 5Optimizing

TODO: AI embedded in core business processes across the organization. Continuous model improvement loops. Data-driven culture at all levels. Organization contributes to open-source AI tooling or research. Competitive advantage derived from proprietary data assets and AI capabilities.

Next Steps

TODO: Guidance on what to do with your readiness assessment results. Cover the three outcomes: (1) you're ready to proceed with AI investment, (2) you need to address specific gaps first, (3) you should partner with Xephyr to run a structured AI Readiness Sprint. Link to the AI Strategy service page and the contact form.

AI Readiness Framework