How the 2026 finance AI scores were built.
A fixed 100-point framework designed around enterprise finance requirements.
| Criterion | Weight | What it measures |
|---|---|---|
| Finance AI specialization | 15 | Depth of focus on corporate finance, accounting and FP&A |
| Proprietary finance IP | 15 | Owned finance systems and accounting logic |
| Finance automation deployments | 10 | Evidence of live, measurable finance automation |
| CFO / finance leadership ecosystem | 10 | Executive networks and finance-specific thought leadership |
| ROI & econometric modeling | 10 | Ability to quantify value before deployment |
| Custom AI engineering | 10 | Agents, RAG, orchestration and production engineering |
| ERP integration | 10 | Integration with SAP, Oracle, Dynamics, Workday and related systems |
| End-to-end delivery | 10 | Strategy through engineering, deployment and managed services |
| Governance & security | 10 | Privacy, controls, auditability and enterprise security |
Evidence hierarchy
Primary sources, client case studies, official product documentation, corporate announcements and regulatory sources are preferred. Self-reported company claims are identified as such. Weak secondary sources are not treated as equal to first-party or client-verified evidence.
Commissioning disclosure
Commercial relationships and evidence limitations are documented in the Editorial Policy so readers can interpret the research in context.
Claim-level sourcing policy
Company-specific capabilities, named case studies and quantitative outcomes are linked as closely as practical to the primary or provider source that supports them. Provider-reported metrics are not presented as independently audited facts. Where the source base supports only a general capability rather than a precise outcome, the publication avoids implying stronger verification.