| name | ai-readiness-assessor |
| description | Assess organizational AI readiness — capabilities, gaps, adoption roadmap for strategic planning. |
| tags | ["planning","strategy","assessment","ai","organizational-readiness"] |
| inherits | planner-base-protocol |
| version | 0.1.0 |
AI Readiness Assessor
JTBD (Jobs To Be Done)
Primary Job
When my organization wants to invest in AI but I can't tell whether the enthusiasm is grounded or whether we're about to buy expensive infrastructure that sits idle,
I want a structured readiness assessment across data, infrastructure, capability, strategy, use cases, and culture,
so I can make investment decisions based on evidence of actual organizational readiness rather than vendor pressure or competitor anxiety.
Secondary Jobs
- When leadership has approved an AI budget but no one has verified whether the organization has the data governance, MLOps maturity, or change management capacity to actually use AI tools, I want those gaps quantified before procurement starts, so the investment doesn't stall at implementation.
- When an AI pilot succeeded in one team but the organization is debating whether to scale it, I want a readiness assessment that distinguishes pilot conditions from enterprise conditions, so the scaling decision accounts for what the pilot had that other teams don't.
Job Layers
- Functional: Assess organizational AI readiness across six dimensions (data maturity, infrastructure, capability, strategy alignment, use case viability, cultural readiness) and produce a prioritized roadmap with sequenced recommendations.
- Emotional: Reduce the anxiety of committing significant AI investment before understanding whether the organization can absorb it — the fear of being the person who championed the AI platform that nobody uses.
- Social: Helps the user present a credible, structured assessment to leadership, board members, and vendors that shows the organization evaluated readiness systematically rather than following hype.
This Skill Is For
- An organization considering significant AI investment that needs to know where the real readiness gaps are before committing budget.
- A leader who needs to distinguish genuine AI readiness from enthusiasm — whether the org has the data, skills, and processes to actually operationalize AI.
- A team that ran a successful AI pilot and needs to assess whether the conditions that made it work exist across the broader organization.
This Skill Is NOT For
- A user with an existing AI strategy or readiness report who needs a quality verdict on it; use
proposal-critic instead.
- A user who needs to plan a specific AI/ML implementation rather than assess organizational readiness; use a domain-specific planner instead.
Paired With
proposal-critic: After the readiness assessment is complete, use it to stress-test the recommendations and roadmap.
stakeholder-report-writer: Use this when the assessment is done but needs to be translated into audience-specific reporting for leadership, board, or technical teams.
Resolution Paths
| User's Situation | What Happens | What They Leave With |
|---|
| Organization wants AI but hasn't assessed readiness | The skill evaluates all six dimensions and scores maturity levels | A prioritized roadmap with sequenced recommendations and quick wins |
| AI budget approved but implementation readiness unknown | The skill identifies specific gaps (data governance, MLOps, change management) blocking implementation | A gap analysis with the specific capabilities to build before procurement |
| Successful pilot needs scaling decision | The skill compares pilot conditions to enterprise conditions | An honest assessment of what scales and what doesn't, with prerequisites |
When to Escalate
- If the user already has an AI strategy document and needs a quality verdict, escalate to
proposal-critic.
- If the user needs the assessment translated into stakeholder-specific reports, escalate to
stakeholder-report-writer.
Purpose
Most organizations that fail at AI adoption don't lack technical capability—they lack readiness across the full organizational ecosystem. This skill conducts a structured assessment across six critical dimensions, producing a prioritized roadmap that prevents costly missteps and accelerates time-to-value.
Why structured assessment matters:
- Organizations that bought expensive ML platforms without data governance infrastructure (wasted 40-60% of budget)
- Teams with pristine data but no MLOps practices (models trained but never deployed)
- Technical readiness achieved but change management failed (adoption rates under 30%)
- Regulatory blindness leading to compliance violations during pilot phases
- Use case selection without ROI validation (high-effort, low-impact projects consuming 18+ months)
This skill aligns business goals, technical reality, organizational capability, and governance requirements into a coherent roadmap with sequenced milestones.
Use_When
- Strategic AI Planning: Organization is deciding whether/how to adopt AI, or scaling from pilot to enterprise
- Investment Validation: Before committing significant budget to AI platforms, vendors, or teams
- Pre-Engagement Assessment: Consulting engagement kickoff to establish baseline and roadmap
- Organizational Transformation: Simultaneous technology and capability building initiatives
- Compliance & Risk: Organizations in regulated industries (healthcare, finance, legal) needing governance frameworks
- Talent & Capability Building: Organizations planning to build internal AI/ML centers of excellence
Do_Not_Use_When
- Assessing readiness for specific projects already approved (use project-scope-analyzer instead)
- Evaluating individual technical tools or platforms (use tool-evaluation-framework)
- Focused solely on data infrastructure without organizational scope (use data-infrastructure-assessor)
- Organizations not ready to act on recommendations (assessment requires commitment to roadmap)
- Technical deep-dives into model performance or ML system design
Why_This_Exists
Common AI Adoption Failures & Prevention
Failure Pattern 1: Technology-First Trap
- Organization: "We'll buy a best-in-class ML platform and hire data scientists"
- Reality: Without data governance, clean data pipelines, and business prioritization, expensive tools sit idle
- Prevention: Assess data readiness first, align technology investments to capability gaps
Failure Pattern 2: Data Infrastructure Without Governance
- Organization: "We have great data lakes and APIs"
- Reality: Without naming standards, lineage tracking, and stewardship models, data scientists spend 80% of time data wrangling
- Prevention: Score governance maturity alongside infrastructure maturity; de-risk before scaling
Failure Pattern 3: Technical Success, Adoption Failure
- Organization: "We trained an amazing churn prediction model"
- Reality: Marketing teams don't trust the model; sales doesn't change workflows; model rots in production
- Prevention: Build use-case validation and change management into the roadmap from discovery
Failure Pattern 4: Isolated Center of Excellence
- Organization: "We hired 5 PhD data scientists"
- Reality: Teams can't collaborate with domain experts; no pathway to push models into production; talent burns out
- Prevention: Assess cross-functional capability and MLOps maturity; identify collaboration dependencies
Failure Pattern 5: Regulatory Blind Spot
- Organization: "We're deploying customer-facing models"
- Reality: Model bias triggers fair lending violations; no audit trail for explainability; GDPR deletion requests fail
- Prevention: Assess governance and compliance posture before pilot; build guardrails into roadmap
Companion_Skills
- stakeholder-report-writer: Convert assessment output into executive summaries for specific stakeholder audiences (C-suite, board, technical team)
- proposal-critic: Validate AI proposals against readiness assessment; identify high-risk initiatives
- data-strategy-planner: Deep-dive into data governance and architecture recommendations from the assessment
- talent-acquisition-advisor: Define hiring profiles and contracting models based on capability gaps
- change-management-designer: Operationalize the culture and change management roadmap
Steps
Phase 1: Organization Context & Goals (Gather)
Inputs to collect:
-
Organization Profile
- Industry vertical and segment (enterprise SaaS, healthcare, retail, financial services, manufacturing, etc.)
- Company size (headcount, revenue)
- Geographic footprint and data residency requirements
- Current technology stack (legacy systems, cloud platforms, custom applications)
- Reporting structure and decision-making authority
-
AI Aspirations & Drivers
- Primary business objectives (revenue growth, cost reduction, risk mitigation, customer experience, competitive positioning)
- Specific use cases under consideration (prioritized or exploratory list)
- Stakeholder appetite for AI (enthusiastic, cautious, skeptical by function)
- Existing AI investments or failed initiatives
- Competitive landscape pressures
-
Constraints & Guardrails
- Budget ceiling for AI programs (Year 1, 3-year)
- Regulatory environment (HIPAA, GDPR, FCA, sector-specific rules)
- Data residency/sovereignty requirements
- Risk tolerance (conservative vs aggressive)
- Timeline expectations (pilot in 6 months vs 18+ months transformation)
Assessment technique: Structured interview with business sponsor, CTO/CIO, and domain leaders. Document as narrative + structured JSON.
Phase 2: Current State Assessment (Score)
Evaluate each of 6 dimensions on a 1-5 maturity scale with evidence. Score reflects actual capability today, not aspirations.
Dimension 1: Data Readiness
Definition: Quality, accessibility, governance, and organizational literacy around data assets
1 - Ad-hoc:
- Data lives in silos (departmental databases, spreadsheets, email)
- No data dictionary or cataloging; teams don't know what data exists
- Data quality issues are frequent and reactively fixed
- No formal data governance; data stewardship is informal
- Data literacy is low; most non-technical stakeholders can't query data
- Evidence: "Data requests take 3+ weeks" / "We discovered duplicate customer records after 6 months"
2 - Managed:
- Centralized data warehouse or lake exists; core business data is accessible
- Basic cataloging effort underway; some metadata tracked
- Data quality rules exist for critical datasets; issues caught in reporting
- Data governance committee meets quarterly; roles assigned informally
- Finance and marketing can run SQL; broader org needs analyst support
- Evidence: "We have a Snowflake data warehouse but no lineage tracking" / "Data governance is in someone's job description"
3 - Standardized:
- Data warehouse/lake with automated pipelines; most operational data flows daily
- Data catalog with 70%+ coverage; lineage tracking for critical pipelines
- Data quality metrics defined and monitored; automated alerts for anomalies
- Formal data stewardship model with owner-per-dataset; governance council reviews policies quarterly
- Self-service BI tools widely adopted; data engineering and analytics teams collaborate on standards
- Evidence: "We have dbt documentation" / "Data quality dashboards show SLAs met 98% of time"
4 - Optimized:
- Real-time and batch pipelines orchestrated; data latency measured and guaranteed
- Data catalog integration with discovery tools; data lineage includes impact analysis
- Data quality platform (Great Expectations, Soda) in use; automated remediation for known issues
- Formal data contracts between producers and consumers; governance enforced in tooling
- Most analysts and business users can self-serve on structured queries; advanced analytics done by specialists
- Evidence: "Data freshness SLAs are <4 hours" / "We have data contracts in place"
5 - Innovating:
- Federated data architecture with real-time integration; data mesh principles applied
- AI-driven data discovery and anomaly detection; automatic lineage and impact recommendations
- Data quality prediction and prevention (modeling data issues before they impact production)
- Data governance automated through metadata platforms; compliance validation built into pipelines
- Entire organization (product, marketing, finance, operations) uses data in decision-making daily
- Evidence: "We use data contracts and federated governance" / "Data quality issues are predicted 80% of the time"
Scoring Evidence Checklist:
- How long do data requests take? (weeks vs days vs hours)
- What % of your data is cataloged and searchable?
- What's your data quality SLA and how often is it met?
- Who owns data governance? Full-time role or add-on responsibility?
- Can non-technical stakeholders access data or do they need analyst translation?
Dimension 2: Technical Infrastructure
Definition: Compute capacity, architecture, MLOps practices, API/integration design, security and privacy controls
1 - Ad-hoc:
- On-premises infrastructure or legacy cloud (single vendor, limited flexibility)
- No ML platform; models trained in notebooks and exported manually
- APIs are few and inconsistent; integrations are point-to-point scripts
- Security is perimeter-based; no data encryption in transit/at rest
- Privacy by default doesn't exist; no PII handling policy
- Evidence: "We train models in Excel/Python locally" / "Models deploy by copying files to a server"
2 - Managed:
- Mix of on-premises and cloud (AWS/Azure/GCP); some modernization started
- ML development tools exist (Jupyter, scikit-learn); no production ML framework
- Some APIs built using frameworks (REST/gRPC); 20-30% of integrations API-first
- Security basics: firewalls, VPNs, some encryption; compliance standards known but not fully implemented
- Privacy by design limited; PII handling is manual and inconsistent
- Evidence: "We use AWS but don't have automated model training" / "We have some APIs but many batch jobs"
3 - Standardized:
- Hybrid cloud strategy defined; cloud-native architecture adopted (containers, microservices)
- MLOps platform in use (SageMaker, Datadog, Kubeflow, or similar); model versioning and experiment tracking
- API-first architecture for 50%+ of integrations; API gateway and standards documented
- Data encryption standard for all new systems; identity and access management (IAM) in place
- Privacy impact assessments required for new data uses; PII masking/anonymization in dev/test environments
- Evidence: "We use Kubernetes and Docker" / "We have a model registry and CI/CD for models"
4 - Optimized:
- Multi-cloud or cloud-agnostic architecture; infrastructure as code and GitOps practices
- MLOps fully automated: model training on schedule, A/B testing, automated retraining, monitoring/alerting
- API-first, 80%+ of integration; API contracts and versioning; gateway with rate limiting and auth
- Zero-trust security model; encryption throughout; secrets management (Vault, AWS Secrets Manager)
- Privacy embedded: PII detection, differential privacy, federated learning experimentation, audit logs for all data access
- Evidence: "Our models retrain on schedule and we A/B test in production" / "All APIs are versioned with OpenAPI specs"
5 - Innovating:
- Autonomous infrastructure management; AI-driven resource optimization and auto-scaling
- MLOps on automated data pipelines; models deployed to edge; continuous learning with human-in-the-loop feedback
- Real-time event-driven APIs; GraphQL and API composition patterns; microgateway architecture
- Decentralized identity and privacy-preserving ML (federated, differential privacy); automated compliance validation
- Privacy-first data architecture; data minimization enforced; regulatory compliance automated
- Evidence: "We deploy models to edge devices" / "We use differential privacy in our ML pipeline"
Scoring Evidence Checklist:
- Where does your compute live (on-prem, single cloud, multi-cloud)?
- How do you train, version, and deploy ML models? (Manual vs automated)
- What % of integrations are API-based?
- Is encryption in place for data in transit and at rest?
- Who can access production data and how is it audited?
Dimension 3: Organizational Capability
Definition: In-house AI/ML talent, cross-functional collaboration, leadership literacy, change management capacity, training programs
1 - Ad-hoc:
- No dedicated AI/ML team; technical staff build models part-time
- Single data scientist or no one; external consultant engaged ad-hoc
- Siloed departments; business and engineering don't collaborate on problems
- Leadership views AI as a technology problem, not a business transformation
- No formal training; learning is self-directed
- Evidence: "Our database admin builds models as a side project" / "We bring in consultants for each project"
2 - Managed:
- 1-2 dedicated data scientists/engineers; mostly hire contractors
- Part-time product-data analytics team; some business analysts
- Cross-functional projects happen; collaboration is informal and reactive
- Leadership believes in AI potential but delegates execution to technical team
- Training offered but not mandatory; YouTube and conferences are primary sources
- Evidence: "We have one senior data scientist" / "We use Upwork for specialized skills"
3 - Standardized:
- 3-5 dedicated AI/ML engineers; mix of internal and contract talent
- Dedicated analytics platform team; data engineers focused on pipeline quality
- Cross-functional product/analytics/engineering teams meet regularly; standard processes for collaboration
- Leadership participates in quarterly AI planning; business sponsorship for projects
- Internal training programs established (lunch-and-learns, courses); certifications encouraged
- Change management plan documented for major AI rollouts
- Evidence: "We have a 4-person ML team and a data engineer" / "We run monthly cross-functional planning meetings"
4 - Optimized:
- 8-12 person AI/ML team; specialized roles (platform engineers, MLOps, domain experts); strategic hiring plan
- Multiple squads aligned to business domains; strong partnership with product and operations
- Formal governance: technical review boards, architecture decisions, standards and best practices documented
- CIO/CTO owns AI roadmap; board-level reporting on AI metrics and ROI
- Formal training program: internal university, vendor certifications, conference attendance budget; internal knowledge sharing forums
- Dedicated change management and organizational development resources
- Evidence: "We have data scientists, ML engineers, and platform engineers with specialized roles" / "We have a formal AI governance board"
5 - Innovating:
- 15+ person AI/ML organization including researchers, platform engineers, domain specialists
- Distributed model with AI embedded in every product squad; centralized platform and research teams
- Continuous talent development; partnerships with universities and research institutions
- CEO/board actively engaged in AI strategy; quarterly business reviews on AI portfolio
- Internal research track alongside product delivery; patents and open source contributions
- Organizational structures redesigned around AI capabilities; change management is embedded in culture
- Evidence: "We hire PhDs and fund research" / "We contribute to open source ML projects"
Scoring Evidence Checklist:
- How many dedicated AI/ML staff do you have (headcount and full-time equivalent)?
- How easily can data scientists, engineers, and business stakeholders collaborate?
- Does leadership actively sponsor AI initiatives or delegate?
- What % of staff participate in formal AI/ML training annually?
- Who is accountable for change management in major AI rollouts?
Dimension 4: Strategy & Governance
Definition: AI strategy aligned to business goals, ethical frameworks, responsible AI policies, regulatory compliance, risk management
1 - Ad-hoc:
- No formal AI strategy; projects initiated by individuals
- No ethics or responsible AI framework; fairness/bias not discussed
- Compliance reactive; only legal escalations trigger governance
- Risk management for AI absent; models deployed without review
- No audit trail or explainability requirements
- Evidence: "We don't have an AI strategy; we do AI projects as they come up" / "We've never discussed model bias"
2 - Managed:
- AI strategy exists but not formally written or communicated; leadership aware but not actively driving
- Basic fairness discussion; awareness that bias could be a problem; no systematic testing
- Compliance known; some policies drafted; audit trail incomplete
- Risk register for AI exists; some model review before production (technical review only)
- Explainability discussed for high-risk models but not enforced
- Evidence: "We have AI goals but no written strategy" / "We've done one bias audit"
3 - Standardized:
- AI strategy written and communicated; aligned to business goals; reviewed annually
- Responsible AI framework documented; fairness, transparency, accountability, and privacy principles defined
- Compliance requirements mapped; policies for GDPR, CCPA, sector-specific rules (HIPAA, FCA) drafted and communicated
- Risk governance: design review, model cards, fairness testing before production; incident response plan
- Audit trail for model decisions (especially high-stakes: lending, healthcare, hiring); explainability standard for regulated models
- Evidence: "We have a documented AI ethics policy" / "Models require fairness testing before deployment"
4 - Optimized:
- AI strategy integrated with business strategy; quarterly business reviews on AI portfolio performance and risk
- Responsible AI maturity model in place; fairness, privacy, security, and explainability tested routinely
- Compliance automated where possible (data discovery, PII handling); audit trails comprehensive; quarterly compliance review
- Risk governance embedded in development: automated fairness checks, model monitoring for drift/bias, guardrails in production
- Explainability and audit trails automatic for all models; customer-facing decisions always explainable on request
- Third-party audits or certifications pursued (AI audits, industry standards)
- Evidence: "We have automated bias detection in our CI/CD" / "We maintain audit trails for all model decisions"
5 - Innovating:
- AI strategy drives competitive advantage; AI organizational transformation underway
- Responsible AI embedded in culture; proactive design for fairness, transparency; external thought leadership
- Compliance preventive; regulatory changes anticipated and built into standards; collaboration with regulators
- Risk anticipation through scenario modeling; adversarial testing; continuous governance improvement
- Explainability by design; models built with interpretability; compliance automation extends to ethics and fairness
- Industry leadership on responsible AI; public commitment and accountability
- Evidence: "We publish our responsible AI principles" / "We participate in AI governance working groups"
Scoring Evidence Checklist:
- Is there a written AI strategy aligned with business goals?
- What responsible AI principles are documented (fairness, privacy, transparency, accountability)?
- What compliance frameworks apply to your industry and which have you addressed?
- What's your process for reviewing models before production?
- Can you explain the reasoning behind high-stakes model decisions?
Dimension 5: Use Case Readiness
Definition: Problem-solution fit, ROI estimation, prioritization rigor, pilot design, success metrics definition
1 - Ad-hoc:
- Use cases identified informally by stakeholders; little validation
- ROI guessed; business impact not quantified
- Priority based on interest or loudest voice; no portfolio management
- Pilots are one-offs; success criteria vague ("see if it works")
- Metrics reactive; chosen after results are in
- Evidence: "We want a chatbot because competitors have one" / "We'll know if the model is good when we see the results"
2 - Managed:
- Use cases documented; basic problem statement written
- ROI estimated roughly (cost of vendor/team vs rough savings)
- Prioritization criteria discussed; not formally scored
- Pilot scope defined; success criteria written but vague
- Some metrics defined upfront; focus on model accuracy
- Evidence: "We want to reduce churn by 5%" / "We'll measure F1 score and AUC"
3 - Standardized:
- Use case template; problem statement, success criteria, and metrics documented
- ROI estimation systematic: baseline cost, expected lift, implementation cost; 3-year model built
- Prioritization matrix: impact × feasibility × alignment scoring; portfolio view of top 10 use cases
- Pilot scope limited; 3-month runway; success criteria include business metrics and adoption
- Metrics balanced: model performance (accuracy, precision, recall), business impact (revenue, cost, time saved), and adoption metrics
- Evidence: "We scored each use case on impact, feasibility, and alignment" / "Pilot success includes business metrics and technical metrics"
4 - Optimized:
- Use case assessment deep: customer research, competitive analysis, build-vs-buy evaluation
- ROI comprehensive: includes implementation cost, ongoing operations, change management, risk discount; sensitivity analysis
- Prioritization sophisticated: considers dependencies, sequencing, and capability building; portfolio balancing (quick wins vs strategic)
- Pilot design rigorous: control group, A/B testing, clear launch criteria, rollback plan
- Metrics comprehensive: technical (model performance), business (revenue, cost, velocity), operational (latency, uptime), and adoption metrics with targets
- Evidence: "We evaluated 3 vendors and 2 build options" / "We run A/B tests for all pilots with control groups"
5 - Innovating:
- Use case discovery continuous; machine learning applied to problem identification
- ROI dynamic; models updated as new data arrives; value realization tracked continuously
- Portfolio optimization automated; machine learning used to recommend next priorities
- Pilot design adaptive; continuous experimentation and iteration; rapid learning
- Metrics real-time and predictive; leading indicators of success; automatic alerts and optimizations
- Evidence: "We use ML to predict which use cases will succeed" / "Metrics update hourly and feed model retraining"
Scoring Evidence Checklist:
- How do you select AI/ML use cases? (Formal scoring or informal?)
- Can you articulate the ROI for each major initiative (baseline, lift, costs)?
- How many AI projects are in your pipeline and how are they prioritized?
- What does a pilot look like? (Timeframe, success criteria, adoption metrics)
- How do you measure success? (Model performance alone or business impact too?)
Dimension 6: Culture & Change Management
Definition: Innovation culture, fear/resistance awareness, communication readiness, stakeholder engagement, learning culture
1 - Ad-hoc:
- Innovation culture weak; change is feared
- Resistance to AI widespread (job loss fears, trust issues, skill anxiety) but not surfaced or addressed
- Communication ad-hoc; AI benefits not clearly articulated
- Stakeholder engagement reactive; affected teams learn late
- Learning culture weak; training is optional and not valued
- Evidence: "People worry AI will replace their jobs" / "We don't talk much about AI unless management brings it up"
2 - Managed:
- Innovation encouraged but inconsistently; some teams embrace change, others skeptical
- Resistance awareness emerging; some fear/anxiety heard but not systematically addressed
- Communication happening; quarterly town halls on AI; some FAQs and resource centers
- Early stakeholder engagement with champions; late involvement of affected users
- Learning available; online courses, but participation low; no budget for conferences
- Evidence: "We have some innovation pilots" / "People are curious but skeptical about AI"
3 - Standardized:
- Innovation culture active; experimentation expected; safe-to-fail frameworks
- Resistance mapping done for major initiatives; concerns documented and addressed
- Communication plan standard: vision, benefits, timeline, skill requirements, role changes clearly stated
- Stakeholder engagement early; working groups form for major initiatives; affected users participate in design
- Learning culture emerging; annual training budget per employee; internal communities of practice
- Change management support: dedicated roles, training, clear ownership
- Evidence: "We run regular hackathons and innovation sprints" / "We have a communication plan and FAQ for each major AI initiative"
4 - Optimized:
- Innovation culture strong; risk-taking normalized; failure is learning opportunity
- Resistance understood deeply through surveys and interviews; proactive mitigation (role expansion, retraining, career paths)
- Communication multilayered: exec messaging, team updates, 1:1 conversations; feedback loops built in
- Stakeholder engagement continuous; steering committees form early; user research and testing integrated
- Learning culture embedded; career development plans include AI/ML skills; mentorship and peer learning active
- Change management integrated with project delivery; dedicated change managers for large initiatives
- Evidence: "We celebrate experiments that fail" / "We've reskilled 20% of the workforce for AI roles"
5 - Innovating:
- Innovation mindset pervasive; continuous experimentation; failure not just tolerated but leveraged
- Resistance prevention through culture building; teams proactively develop new skills; career ladders built
- Communication transparent and two-way; employees co-create vision and roadmap
- Stakeholder engagement participatory; affected teams design solutions alongside AI/ML experts
- Learning ecosystem comprehensive; internal academy, external partnerships, research projects; continuous skill building
- Change management predictive; culture shifts anticipated and designed for; organizational structure evolves with strategy
- Evidence: "Employees drive half the AI use case ideas" / "We have internal AI research projects"
Scoring Evidence Checklist:
- How are new ideas and innovations encouraged or discouraged in your organization?
- Are people worried about job security due to AI? How are those concerns being addressed?
- How do employees learn about major AI initiatives and changes coming?
- How involved are affected teams in designing AI solutions before rollout?
- Is there budget and time for learning and skill development in AI/ML?
Phase 3: Gap Analysis (Analyze)
For each dimension, compare current state (Phase 2) to required state for their priority use cases and business goals.
Gap Analysis Framework:
-
Define Target Maturity: For each use case (e.g., "predictive churn model"), what maturity level is required?
- Data-heavy personalization: Data Readiness 4+, Use Case Readiness 4+
- Regulatory compliance (healthcare ML): Strategy & Governance 4+
- Customer-facing recommendation engine: Culture 3+, Organizational Capability 3+
-
Calculate Gaps: Where current maturity < required, define the gap
- Example: Data Readiness 2 → 4 required (gap of 2 levels)
- Example: Culture & Change Management 2 → 3 required (gap of 1 level)
-
Dependency Analysis: Which gaps must be addressed before others?
- Data governance (Dimension 1) enables technical infrastructure (Dimension 2)
- Organizational capability (Dimension 3) enables governance (Dimension 4)
- Strategy & governance (Dimension 4) informs use case prioritization (Dimension 5)
-
Risk Register: For each significant gap, identify:
- What could go wrong if we skip this?
- Timeline to close the gap
- Cost and resource requirements
- Dependencies and blockers
Phase 4: Roadmap Design (Plan)
Create a phased roadmap with sequenced recommendations.
Roadmap Structure:
Quick Wins (0-3 months)
- High impact, low effort initiatives that build momentum and demonstrate progress
- Examples: Establish data governance committee, hire first ML engineer, launch AI literacy program, audit compliance readiness
- Builds confidence and funds later initiatives
Foundation Building (3-9 months)
- Address critical gaps in data, infrastructure, or organizational capability
- Examples: Implement data cataloging, establish MLOps practices, build initial cross-functional teams, document AI strategy
- These enable larger initiatives
Scaling Phase (9-18 months)
- Deploy multiple use cases in parallel; operationalize governance; embed AI in business processes
- Examples: Production ML systems, expanded team, automated compliance checks, advanced analytics platform
- Return on investment begins
Optimization Phase (18+ months)
- Continuous improvement; advanced capabilities; organizational transformation
- Examples: Real-time ML, edge deployment, research partnerships, customer-facing AI products
- Competitive advantage solidified
Recommendations by Dimension:
- Data Readiness: Implement data cataloging → data governance policy → data quality platform → federated data architecture
- Technical Infrastructure: Cloud modernization → containerization and APIs → MLOps platform → edge/real-time capabilities
- Organizational Capability: Hire core ML team → build cross-functional squads → establish centers of excellence → embed AI in every team
- Strategy & Governance: Write AI strategy → responsible AI framework → compliance policies → automated governance
- Use Case Readiness: Document use cases and prioritize → run pilots → scale successful pilots → continuous use case discovery
- Culture & Change Management: Launch AI literacy program → establish communities of practice → reskill and rehire → transform organizational culture
Phase 5: Implementation Planning (Execute)
Create detailed implementation plan with timelines, budgets, resource needs, and success metrics.
Implementation Plan Components:
- Timeline: Quarters/milestones with clear start/end dates and deliverables
- Budget: Cost ranges for each phase (personnel, tools, training, consulting)
- Resource Plan: Headcount and skills needed (data engineers, ML engineers, data scientists, product managers, change managers)
- Success Metrics: How will we know the roadmap is working? (adoption rates, time-to-value, quality improvements, employee engagement)
- Review Checkpoints: When will we assess progress and adjust? (monthly, quarterly, semi-annually)
- Risk Mitigation: For each major risk from the risk register, define how it will be managed
- Sponsorship & Governance: Who owns each phase? Who makes decisions? Escalation paths
Full_AI_Readiness_Protocol
The skill implements a 5-phase planning protocol embedded in the prompt that will be invoked by the agent:
PHASE 1: ORGANIZATION CONTEXT & GOALS
├─ Gather industry, company size, tech stack
├─ Understand AI aspirations and drivers
├─ Document constraints (budget, regulatory, risk tolerance)
└─ Output: Context narrative + structured data
PHASE 2: CURRENT STATE ASSESSMENT
├─ Score Data Readiness (1-5) with evidence
├─ Score Technical Infrastructure (1-5) with evidence
├─ Score Organizational Capability (1-5) with evidence
├─ Score Strategy & Governance (1-5) with evidence
├─ Score Use Case Readiness (1-5) with evidence
├─ Score Culture & Change Management (1-5) with evidence
└─ Output: Maturity Scorecard + spider diagram data
PHASE 3: GAP ANALYSIS
├─ Define target maturity for each dimension
├─ Calculate gaps (current vs target)
├─ Analyze dependencies
├─ Build risk register
└─ Output: Gap summary + risk register
PHASE 4: ROADMAP DESIGN
├─ Quick Wins (0-3 months)
├─ Foundation Building (3-9 months)
├─ Scaling Phase (9-18 months)
├─ Optimization Phase (18+ months)
└─ Output: Sequenced recommendations by dimension
PHASE 5: IMPLEMENTATION PLANNING
├─ Timeline with milestones
├─ Budget estimates (personnel, tools, consulting)
├─ Resource plan (headcount, skills, hiring)
├─ Success metrics and review checkpoints
├─ Risk mitigation strategies
└─ Output: Detailed implementation roadmap
Tool_Usage
This skill is read-only and planning-focused. It does not:
- Modify documents or code
- Execute changes to systems
- Make purchasing decisions or vendor selections
- Commit budget or resources
It does:
- Conduct interviews and assessments
- Analyze organizational documents (tech stack, org charts, strategic plans)
- Synthesize input into structured recommendations
- Output detailed roadmaps ready for executive review and decision-making
Ideal workflow:
- Use this skill to assess readiness and create roadmap
- Use stakeholder-report-writer to tailor output for board, C-suite, or technical teams
- Use proposal-critic to validate AI proposals against readiness findings
- Use change-management-designer to operationalize the roadmap
Examples
Example 1: Mid-Market SaaS Company (Series B)
Organization Profile:
- 80 employees, $15M ARR, 5-year-old SaaS platform
- Current tech stack: Node.js/React frontend, PostgreSQL database, basic analytics
- Industry: HR technology (recruiting/onboarding)
AI Aspirations:
- Build resume screening and candidate matching features (core product differentiation)
- Reduce customer support volume through smart ticketing and routing
- Predictive analytics for customer churn and upsell opportunities
Assessment Results (from Phase 2):
Data Readiness : 2/5 (Managed)
- PostgreSQL is robust but no data warehouse
- No data governance; analytics team manually builds reports
Technical Infrastructure : 2/5 (Managed)
- AWS-based but no containerization or MLOps
- APIs exist but not comprehensive; integrations are custom
Organizational Capability : 1/5 (Ad-hoc)
- No dedicated data/ML team; VP Engineering manages analytics part-time
- Single analyst; no cross-functional collaboration structure
Strategy & Governance : 1/5 (Ad-hoc)
- No formal AI strategy
- Fairness/bias not discussed; compliance aware but reactive
Use Case Readiness : 2/5 (Managed)
- Use cases identified but not formally prioritized
- ROI guessed for resume screening feature
Culture & Change Management : 2/5 (Managed)
- Engineering team is innovative; product team skeptical
- No formal communication about AI plans
Gap Analysis for Resume Screening Use Case:
- Target: Data 3, Technical 3, Capability 3, Governance 2, Use Case 3, Culture 2
- Gaps: Data (+1), Technical (+1), Capability (+2), Governance (+1), Use Case (+1)
Roadmap Excerpt:
Quick Wins (0-3 months):
- Hire first ML engineer (part-time contractor → full-time)
- Build basic data pipeline from Postgres to Redshift
- Document AI strategy and responsible AI principles
Foundation (3-9 months):
- Implement data governance (data catalog, naming standards)
- Build resume screening MVP using existing candidate data
- Create cross-functional product-data-ML team
Scaling (9-18 months):
- Deploy resume screening to production; A/B test with customers
- Build support ticket routing ML model in parallel
- Expand ML team; establish data engineering role
Example 2: Enterprise Healthcare Organization
Organization Profile:
- 500+ employees, healthcare provider, 3 hospital systems
- Current tech stack: Legacy EHR system, multiple data silos, limited cloud use
- Regulatory environment: HIPAA, state licensing requirements
AI Aspirations:
- Predictive ICU admissions (identify high-risk patients early)
- Clinical decision support (drug interaction alerts, diagnosis suggestions)
- Administrative ML (claims processing, billing optimization)
Assessment Results (Phase 2):
Data Readiness : 2/5 (Managed)
- Data scattered across EHR, pharmacy, lab systems
- No centralized patient data repository
- Data quality issues common; limited governance
Technical Infrastructure : 1/5 (Ad-hoc)
- On-premises infrastructure; limited cloud
- No APIs; system integrations are batch files
- Security and privacy basics in place; encryption partial
Organizational Capability : 2/5 (Managed)
- One bioinformaticist; no data engineering or ML team
- Clinical staff and IT don't collaborate on requirements
Strategy & Governance : 2/5 (Managed)
- AI strategy not formal; regulatory compliance is focus
- Fairness in medicine discussed informally
- Risk management exists (medical errors) but not AI-specific
Use Case Readiness : 2/5 (Managed)
- Multiple use cases identified; ROI unclear
- No prioritization; clinical leadership wants everything
Culture & Change Management : 2/5 (Managed)
- Clinicians fear automation will replace judgment
- Resistance to change is cultural; training is minimal
Gap Analysis for ICU Admissions Prediction:
- Target: Data 4, Technical 3, Capability 3, Governance 4 (regulatory), Use Case 3, Culture 3
- Gaps: All dimensions have gaps (+2 to +3)
- Unique risks: Patient privacy, model bias in healthcare, regulatory approval before deployment
Roadmap Excerpt:
Quick Wins (0-3 months):
- Form clinical-IT steering committee
- Conduct bias/fairness audit on historical ICU data
- Document AI governance and responsible AI principles for healthcare
- Hire clinical informatics consultant
Foundation (3-9 months):
- Establish Health Information Exchange (HIE); begin data consolidation
- Launch AI literacy program for clinical staff (focus on explainability and trust)
- Build robust data governance with patient privacy as foundation
- Develop regulatory compliance and explainability requirements
Scaling (9-18 months):
- Deploy ICU prediction model with clinical review workflow
- Conduct fairness validation across patient populations
- Pilot administrative ML models
- Expand clinical informatics and data engineering teams
Notes
- Assessment is not one-time: Organizational readiness evolves. Re-assess annually or when major strategic changes occur.
- Maturity levels are relative: A score of 2 for a healthcare org might be acceptable; same score for a tech company might be a critical gap.
- Dependencies matter: Trying to scale ML models (Dimension 2) without data governance (Dimension 1) typically fails. The roadmap sequence honors these dependencies.
- Culture is hardest to change: Technical capability can be hired or built in 6-12 months. Culture and change management require 18-24+ months.
- Governance is not optional: Any regulated industry (healthcare, finance, legal) requires strong governance before scaling. Strategy & Governance should never be < 3 for production systems.
- Use case selection drives everything: Pick the right first use case and momentum builds. Pick wrong and the AI initiative loses credibility.
- Budget varies widely: A mid-market SaaS company might spend $500K-$1M in Year 1 (team + tools). An enterprise might spend $2-5M. Healthcare and finance often spend more due to compliance and integration complexity.
- Timeline is realistically 18-24 months from strategy to scaled operations. Anything faster typically sacrifices governance or quality.
Related Resources
- AI/ML Maturity Models: NIST AI Risk Management Framework, SEI CMMI for ML, Gartner AI Maturity Model
- Responsible AI Frameworks: Partnership on AI Responsible AI Practices, IEEE Ethically Aligned Design
- MLOps References: ML Model Ops Manifesto, Google's Rules of ML, MLOps.community
- Change Management: Prosci ADKAR Model, Kotter's 8-Step Process, McKinsey Change Management Framework
- Data Governance: DAMA Data Management Body of Knowledge, Gartner Data Governance