RACI Chart and Governance Template
To ensure clarity in roles and accountability throughout the AI adoption program, creating RACI matrices (Responsible, Accountable, Consulted, Informed) for key processes is recommended. Below we provide sample RACI charts for two levels:
- AI Adoption Program Governance (macro level) – who owns the strategy, funding, risk oversight, etc.
- Individual AI Project Execution (micro level) – who does what in a given project from Ideate to Activate.
A. AI Program Governance RACI
| Activity / Decision | Responsible (R) | Accountable (A) | Consulted (C) | Informed (I) |
| Define AI Strategy & Vision | Head of AI/Analytics CoE | CIO/CTO (Executive Sponsor) | CEO, Business Unit Heads | Board of Directors (high-level brief) |
| Prioritize AI project portfolio | AI Steering Committee PMO | Executive Sponsor (e.g. CIO) | CFO (for budget), BU Heads (for alignment) | All BU Heads, Program stakeholders |
| Establish AI Ethical Guidelines | Chief Data Officer (CDO) / AI Ethics Officer | General Counsel | Risk Management, External Ethics Advisor | All employees (policy roll-out) |
| Approve project funding (gate between phases) | AI Steering Committee | CFO (for funding) | CIO, BU Head for project, Risk/Compliance | Project Team Leads |
| Monitor program progress (KPIs) | AI Program Management Office (PMO) | CIO/CTO | BU Heads (progress in their domain), CFO (benefit tracking) | CEO, Board (via quarterly reports) |
| AI Risk & Compliance oversight | AI Governance Board (subset focusing on risk) | Chief Risk Officer (CRO) or General Counsel | IT Security, Data Privacy Officer, External regulators (if needed) | Board Risk Committee (summary) |
| Change management & training plan (enterprise) | HR Learning & Dev Lead | CHRO (Chief HR Officer) | CIO/CTO, BU Heads (for scheduling, content input) | All employees (announcement of training) |
| Technology platform selection (for AI) | Head of Data Engineering/Architecture | CTO | CISO (security input), Procurement (costs), AI CoE | AI project teams (to use standards) |
| Vendor/Partnership decisions (e.g. OpenAI API usage) | AI CoE Lead / Procurement | CIO | Legal (contract), IT Security (vetting), BU stakeholder | CFO (for major contracts), Steering Comm. |
| Communication of AI program updates | AI Program Communications Lead | CIO or CDO | Corporate Communications (messaging), BU Reps (success stories) | All employees, sometimes public (press releases if external) |
In the above:
- The Executive Sponsor (CIO/CTO) is Accountable for the overall program – ensuring it meets objectives.
- A Steering Committee (possibly led by the CIO and including BU heads, CDO, CFO, etc.) collectively is consulted on priorities, but one person holds A.
- A Chief Data Officer or AI CoE head is often R for many day-to-day actions (they run the program management).
- Risk oversight might have its own subcommittee; here CRO is A for risk decisions like approving use of a sensitive AI application.
- Importantly, Business Unit Heads are consulted on things that impact their domain (ensuring alignment and adoption).
- The Board is mainly informed at major milestones or for overall risk (they won’t manage directly but need to know status and issues, especially at high maturity where AI is strategic or risky).
B. Individual AI Project RACI
(for a single use case):
(This can vary per project, but a generic template:)
| Phase / Task | Responsible (R) | Accountable (A) | Consulted (C) | Informed (I) |
| Use Case Definition (Ideate) | Business Analyst or BU Sponsor (for that use case) | Business Unit Head (Owner of process) | AI CoE Rep (to ensure fits strategy), Data Scientist (feasibility) | AI Program PMO |
| Data Gathering & Preparation | Data Engineer, Data Scientist | AI Project Manager (if assigned) or Tech Lead | Data Owner (of source systems), IT (for access) | BU Sponsor (progress update) |
| Model Development (Innovate) | Data Scientist / ML Engineer | AI Tech Lead (could be data science manager) | Domain SME (ensure outputs make sense), AI CoE (best practices) | BU users (if needed for testing) |
| Pilot Testing Execution | AI Project Manager | Business Unit Head (if process impact) | Pilot Users, Risk Manager (monitor outcomes) | Steering Committee (pilot start/end) |
| Results Evaluation | AI Project Manager + Data Scientist (analysis) | BU Sponsor (for business results) | Finance (validate value calc), Risk (issues) | Steering Comm (detailed in stage gate) |
| Go/No-Go Decision (Gate) | (N/A – decision by committee) | Steering Committee Chair (CIO or delegate) | BU Head, Risk, Compliance (present their stance) | Project team (told decision) |
| Deployment Planning (Activate) | IT Project Manager (deployment) | Business Unit Head (accepting into ops) | AI CoE (for support), IT Ops, Change Management | All pilot users (upcoming changes) |
| Production Deployment Setup | DevOps/MLOps Engineer, IT Ops | IT Application Owner or AI CoE lead (for platform) | Security (for final review), DBA (database changes) | BU Process Owner (when live) |
| User Training (for that solution) | Training Lead (could be from BU or CoE) | Business Unit Head | AI Project Manager (content), HR L&D (if needed) | All End-Users (they attend training) |
| Post-Launch Monitoring | AI Ops/MLOps Engineer | Process Owner or BU Manager (accountable for outcomes) | AI CoE (monitor aggregate trends), Risk Officer (if high-risk model) | Steering Comm (in periodic review) |
| Ongoing Model Maintenance | Data Scientist or AI CoE model owner | BU Process Owner (or CoE if central) | IT Ops (for deployment), Risk (re-validate if changes) | Steering Comm (if major updates) |
In this:
- The Business Unit Head/Process Owner is crucially accountable because they “own” the business outcome. For instance, if this is a marketing AI, the CMO or marketing director is accountable that it achieves marketing outcomes.
- The AI technical lead/CoE is responsible for technical parts.
- AI Project Manager (if one exists) coordinates and might be accountable for timeline/budget.
- Risk/Compliance are consulted especially at go-live and evaluation.
- Users are consulted in design and informed/trained for changes.
- After deployment, the business owner remains accountable for using the AI properly to get results, while IT/AI teams remain responsible for keeping it running well.
These templates can be modified based on organization structure (for example, in some cases the CDO might own deployment rather than BU if it’s a centralized approach, but then handover to BU is needed).
Having clear RACI avoids the common confusion like “Who is responsible if the AI makes a wrong decision causing an error?” Under our matrix: The process owner is accountable for process outcomes, but the AI Ops team is responsible for investigating model failure and fixing it, with risk consulted. This clarity ensures quick action and no finger-pointing.