Finance AI Maturity Assessment
The maturity model in this chapter is an illustrative Oracle framework for CoE planning and discussion. It is not a product entitlement model, implementation requirement, certification, or regulatory classification.
For each finance sub-function, score the six dimensions from 1 to 5 using documented evidence, and record the rationale and owner for each score. Report the per-dimension profile and the lowest score as the primary constraint; an average must not obscure a weak Governance or other critical dimension. CoEs should tailor the dimensions, evidence, thresholds, and target stages to their operating model and risk profile.
The stage descriptions and sub-function examples below are illustrative, not product commitments or prescribed sequencing.
Maturity Stages
| Stage | Name | Defining Characteristic |
|---|---|---|
| 1 | Foundational | No coordinated AI strategy; isolated, individual experimentation |
| 2 | Emerging | Sponsored pilots under way; business cases defined; ROI tracking informal |
| 3 | Operational | AI embedded in production finance processes; defined ownership; explainable outputs |
| 4 | Scaled | AI deployed across all finance sub-functions; ROI tracked at enterprise level |
| 5 | Transformational | Finance operating model redesigned around AI; CFO is enterprise AI strategist |
Assessment Dimensions
| Dimension | Definition | Primary Failure Mode When Absent |
|---|---|---|
| Strategy & Vision | Formal AI strategy, CFO sponsorship, use case portfolio management, and ROI tracking | AI remains a collection of disconnected experiments with no shared direction or accountability |
| Data & Infrastructure | Data platform maturity, integration across ERP, CRM, and planning systems, and real-time data availability | Data preparation overhead stalls AI projects before they reach production |
| Technology & Tools | Embedded versus bolt-on AI architecture, Oracle EPM AI feature deployment, ML model production readiness, and operational resilience procedures | Pilots succeed in test environments but fail to reach or sustain production |
| Talent & Culture | AI literacy across finance staff, no-code AI tool ownership, reduction of data science dependency, and CoE capability | AI capability concentrates in IT or the CoE and does not reach Finance process owners |
| Governance & Risk | Explainability standards for AI outputs, model validation, vendor AI contract governance, regulatory alignment, and bias controls | AI outputs cannot be explained to auditors; regulatory obligations go unaddressed |
| Change Management & Value | Finance process documentation completeness, change program structure, AI ROI measurement, and value attribution | AI is applied to undocumented or inconsistent processes and reflects their disorder at scale |
Sub-function View
Stage 3 (Operational) is the illustrative CoE milestone at which selected AI-generated insights and controlled decision-support workflows operate in production with defined ownership, monitoring, and human review. The examples below show a possible progression by finance sub-function; they do not indicate product availability, entitlements, or required sequencing.
| Sub-function | Stage 1 | Stage 2 | Stage 3 ★ | Stage 4 | Stage 5 |
|---|---|---|---|---|---|
| FP&A | Manual budgeting; static annual plans; ad hoc variance analysis | Self-service BI dashboards; driver-based modeling pilots | ML rolling forecasts; variance narratives; real-time actuals vs plan; strategic modeling; AI-generated insights | Predictive scenario modeling; AI-generated board packs; continuous planning platform | Autonomous forecast updates; AI-driven M&A modeling; self-correcting planning models |
| Controllership | Manual journal entries; spreadsheet close; high error rates | RPA pilots; Enterprise Journals rollout; digital close checklist | Automated reconciliation; AI anomaly detection; closed-loop decision systems; AI-generated insights | Continuous accounting; autonomous account matching; AI-flagged exceptions only | Virtual close near real-time; AI-owned routine transactions; humans approve exceptions |
| Treasury | Manual cash positioning; static liquidity reports | Cash pooling automation; TMS deployment | ML cash forecasting; real-time liquidity dashboard; AI-generated insights | AI counterparty risk monitoring; autonomous liquidity optimization | Fully autonomous treasury operations; AI-led capital allocation |
| Tax | Manual provisions; reactive compliance; spreadsheet transfer pricing | Tax software deployment; structured CbCR data | AI-assisted provision; automated CbCR filings; NLP contract review; AI-generated insights | Predictive tax risk modeling; automated transfer pricing documentation | Proactive tax planning AI; autonomous compliance filing |
| Internal Audit | Annual plan; manual sampling; paper findings management | Data analytics in audit; risk-based plan; digital working papers | AI anomaly testing; automated controls monitoring; continuous risk signals; AI-generated insights | 100% transaction coverage; predictive risk scoring; AI-drafted findings | Continuous autonomous audit; human auditors focus on judgment |
Interpreting the Assessment Result
The assessment identifies the lowest-scoring dimension. That dimension — not the highest — determines the realistic advancement ceiling.
Common patterns:
- Strategy & Vision at Stage 1 or 2: No AI budget or executive sponsor exists. Pilot selection is premature.
- Data & Infrastructure at Stage 1 or 2: Data preparation overhead will stall any pilot before it reaches production. Data platform work must precede AI deployment.
- Governance & Risk below Stage 3: Explainability requirements and vendor AI contract governance are not in place. This is a regulatory non-compliance risk for organizations in scope for the EU AI Act.
- Change Management & Value below Stage 2: Finance processes are undocumented or inconsistent. AI applied to these processes reflects their disorder.