An AI Maturity Framework

Level 1: Crawl

• Are individual employees experimenting with AI tools on their own initiative?

• Has leadership acknowledged AI as a topic of interest or potential?

• Are there ad-hoc discussions about AI in leadership meetings?

Level 2: Walk

• Has leadership defined formal AI objectives and priorities?

• Is there a designated executive sponsor or AI steering committee?

• Has the organization allocated budget specifically for AI initiatives?

• Are AI goals included in strategic planning documents?

Level 3: Run

• Does leadership have a comprehensive AI strategy with clear ROI targets?

• Are AI objectives integrated into departmental and individual performance goals?

• Is there regular executive review of AI initiative progress and outcomes?

• Has leadership championed organization-wide AI adoption?

Level 4: Fly

• Does the board regularly review AI strategy and competitive positioning?

• Are AI capabilities central to the organization's competitive advantage?

• Is there a Chief AI Officer or equivalent C-level position?

• Does leadership actively shape industry AI standards and best practices?

Level 1: Crawl

• Are employees using publicly available or manually gathered data?

• Is data primarily stored in siloed spreadsheets or local files?

• Is data quality managed informally or ad-hoc?

Level 2: Walk

• Is there a centralized data warehouse or lake?

• Are data quality standards documented and monitored?

• Is data lineage tracked for critical datasets?

• Are there established data governance policies?

Level 3: Run

• Is real-time or near-real-time data available for AI applications?

• Are data pipelines automated with monitoring and alerting?

• Is metadata comprehensively managed across all data assets?

• Are data quality metrics consistently above 95% for critical datasets?

Level 4: Fly

• Do automated systems continuously validate and improve data quality?

• Is synthetic data generation used to augment training datasets?

• Are data pipelines self-healing with automatic anomaly detection?

• Is data provenance fully tracked with automated compliance verification?

Level 1: Crawl

• Are employees using consumer-grade AI tools (ChatGPT, Claude, etc.)?

• Is AI experimentation happening on personal devices or accounts?

• Is there no formal AI infrastructure in place?

Level 2: Walk

• Has the organization deployed enterprise AI platforms (Microsoft Copilot, etc.)?

• Is there secure API access to AI models for development teams?

• Are development and production environments separated?

• Is there basic version control for AI-related code?

Level 3: Run

• Is there a complete MLOps platform with CI/CD pipelines?

• Are model training, testing, and deployment automated?

• Is infrastructure scalable with cloud or hybrid capabilities?

• Are there established sandboxes for experimentation?

Level 4: Fly

• Are AI agents deployed on auto-scaling infrastructure?

• Is there a comprehensive AI orchestration platform managing multiple agents?

• Are models automatically retrained based on performance degradation?

• Is infrastructure self-optimizing for cost and performance?

Level 1: Crawl

• Is AI usage largely untracked or unmonitored?

• Are there no formal logs of AI tool usage?

• Is success measured anecdotally if at all?

Level 2: Walk

• Are basic usage metrics collected (number of queries, users, etc.)?

• Is there logging of AI system inputs and outputs?

• Are cost and performance metrics tracked in dashboards?

• Are there alerts for system failures or anomalies?

Level 3: Run

• Is model performance continuously monitored against KPIs?

• Are comprehensive audit trails maintained for compliance?

• Is there real-time monitoring of model drift and data quality?

• Are user satisfaction and business impact metrics tracked?

Level 4: Fly

• Do monitoring systems automatically trigger remediation actions?

• Is there predictive alerting based on trend analysis?

• Are agents self-monitoring with automatic performance optimization?

• Is there comprehensive observability across the entire AI agent ecosystem?

Level 1: Crawl

• Are there no formal AI usage policies or guidelines?

• Is risk management for AI informal or non-existent?

• Are employees unaware of potential AI-related risks?

Level 2: Walk

• Are acceptable use policies documented and communicated?

• Is there a risk assessment framework for AI projects?

• Are there approval processes for new AI initiatives?

• Is regulatory compliance reviewed for AI applications?

Level 3: Run

• Is there a formal AI governance board with regular meetings?

• Are all AI projects subject to ethical review and risk assessment?

• Is IP protection systematically managed for AI-generated content?

• Are third-party AI vendors assessed for compliance and risk?

Level 4: Fly

• Are governance policies automatically enforced through technical controls?

• Is there continuous compliance monitoring with automated reporting?

• Does the organization contribute to industry AI governance standards?

• Are AI agents subject to automated ethical guardrails and override mechanisms?

Level 1: Crawl

• Are employees potentially sharing sensitive data with public AI tools?

• Is there no awareness of AI-specific security threats?

• Are AI tools used without security review?

Level 2: Walk

• Are prompt injection and data leakage risks understood and documented?

• Is sensitive data prohibited from being shared with external AI tools?

• Are enterprise AI tools deployed with security configurations?

• Is there basic training on AI security risks?

Level 3: Run

• Are AI assistants deployed with comprehensive security controls?

• Is there continuous monitoring for AI-specific attack vectors?

• Are security assessments mandatory for all AI implementations?

• Is AI being used to enhance cybersecurity capabilities?

Level 4: Fly

• Do AI agents operate within zero-trust security architectures?

• Are AI-powered security tools autonomously detecting and responding to threats?

• Is there real-time threat intelligence specifically for AI systems?

• Are agents secured with hardware-based trust and attestation?

Level 1: Crawl

• Are individuals using AI tools without coordination?

• Is there no dedicated AI talent or roles?

• Is AI expertise limited to a few enthusiasts?

Level 2: Walk

• Has the organization hired or designated AI specialists?

• Are there cross-functional AI working groups that meet regularly?

• Is there a forum for sharing AI knowledge and use cases?

• Are job descriptions updated to include AI-related responsibilities?

Level 3: Run

• Is there a dedicated AI team or center of excellence?

• Are AI champions embedded across business units?

• Is there a clear career path for AI roles?

• Are retention strategies in place for AI talent?

Level 4: Fly

• Is AI expertise distributed throughout the organization?

• Are teams structured around AI agent development and orchestration?

• Is there a talent pipeline through academic partnerships and internships?

• Does organizational design facilitate rapid AI innovation and deployment?

Level 1: Crawl

• Is AI training limited to individual self-learning?

• Are employees discovering AI tools on their own?

• Is there no formal AI education program?

Level 2: Walk

• Is there firm-wide AI literacy training for all employees?

• Are role-specific AI training programs developed?

• Is there documentation of AI best practices and guidelines?

• Are employees aware of available enterprise AI tools?

Level 3: Run

• Is there continuous learning with regular AI training updates?

• Are employees certified in AI tool usage for their roles?

• Is there a knowledge management system for AI practices?

• Are advanced courses available for power users and developers?

Level 4: Fly

• Are external certifications and advanced degrees supported?

• Is there a formal AI academy or university partnership?

• Do employees contribute to AI research and publications?

• Is training personalized based on role, skill level, and usage patterns?

Level 1: Crawl

• Are AI applications limited to personal productivity?

• Is AI use opportunistic without clear business objectives?

• Are success metrics undefined or not tracked?

Level 2: Walk

• Are use cases identified and prioritized by business function?

• Is there a documented methodology for selecting AI projects?

• Are pilot projects launched with defined success criteria?

• Is ROI tracked for AI initiatives?

Level 3: Run

• Are AI assistants deployed across multiple business functions?

• Is there a portfolio approach to managing AI projects?

• Are use cases regularly evaluated and optimized?

• Is learning from projects systematically captured and applied?

Level 4: Fly

• Are autonomous agents handling end-to-end business processes?

• Is there continuous identification of new automation opportunities?

• Are agents self-optimizing based on performance data?

• Is AI integrated into every major business function with measurable impact?

Level 1: Crawl

• Is analytics limited to using pre-built AI tools?

• Is there no custom AI model development?

• Are capabilities limited to prompting existing models?

Level 2: Walk

• Are analytics teams experimenting with model fine-tuning?

• Is there basic prompt engineering and RAG implementation?

• Are pre-built models integrated into business applications?

• Is there a development environment for AI experimentation?

Level 3: Run

• Are custom models developed and deployed for specific use cases?

• Is there a complete MLOps pipeline from development to production?

• Are A/B testing and champion/challenger models implemented?

• Is model performance continuously evaluated and improved?

Level 4: Fly

• Are AI agents autonomously developing and deploying models?

• Is there automated feature engineering and model selection?

• Do systems automatically optimize across multiple objectives?

• Is there a self-improving AI development pipeline?

Level 1: Crawl

• Are vendor relationships limited to consumer SaaS subscriptions?

• Is there no formal engagement with AI vendors or partners?

• Is external AI expertise not being leveraged?

Level 2: Walk

• Are enterprise agreements established with major AI providers?

• Is the organization participating in industry AI forums?

• Are consultants or implementation partners engaged for specific projects?

• Is there evaluation of specialized AI vendors by use case?

Level 3: Run

• Are strategic partnerships established with key technology providers?

• Is the organization involved in industry consortia and working groups?

• Are academic partnerships in place for research collaboration?

• Are data partnerships established to enhance AI capabilities?

Level 4: Fly

• Is the organization co-developing AI solutions with partners?

• Are there joint ventures or investments in AI startups?

• Does the organization influence partner roadmaps and standards?

• Is there an ecosystem of partners supporting the AI agent infrastructure?

Level 1: Crawl

• Are AI tools used informally without process integration?

• Are workflows unchanged despite AI availability?

• Is there no documentation of how AI fits into processes?

Level 2: Walk

• Are business workflows documented to identify AI opportunities?

• Is there a change management process for AI implementations?

• Are employees trained on new AI-enhanced processes?

• Are processes updated to incorporate enterprise AI tools?

Level 3: Run

• Are workflows optimized around AI assistant capabilities?

• Is there continuous process improvement driven by AI insights?

• Are standard operating procedures consistently followed?

• Is change management sophisticated with user adoption tracking?

Level 4: Fly

• Are processes fully automated with AI agents handling exceptions?

• Is there dynamic process optimization based on real-time data?

• Are agents autonomously improving workflows?

• Is the organization constantly innovating processes using AI capabilities?