Why CFOs Need to Bridge Finance and Tech to Realize AI's Full Value
AI is moving fast, but finance is moving careful. That gap is widening inside large enterprises-especially between CFOs and their CIO/CTO peers. The result: siloed projects, duplicated spend, and higher compliance risk than anyone wants.
New data from EY's 2025 Technology Risk Pulse Survey shows the split clearly. Finance leaders are prioritizing AI, but tech leaders are pushing harder and broader. Without a shared playbook, ROI slows and risk multiplies.
The gap by the numbers
- AI integration is a top priority for 56% of CFOs vs. 70-72% of CIOs/CTOs over the next 2-4 years.
- 77% of CFOs focus AI on financial reporting; 83% of CIOs prioritize IT infrastructure.
- Governance tops the list: 81% of executives rate SOX and ICFR as very or extremely important, 78% cite SOC reporting as critical to audit readiness-and 90% of CFOs rank SOC reporting as a high priority.
The message: finance wants stronger oversight before pushing automation further. That caution is rational-but it can't become a bottleneck. CFOs can lead by aligning investment, risk, and delivery with tech to scale the right use cases at the right speed.
What "bridging" looks like in practice
- Set shared principles: Agree on risk appetite, control-by-design, data boundaries, auditability, and vendor criteria. Put it in writing and use it to greenlight or halt use cases.
- Run a joint portfolio: One backlog owned by finance and tech. Rank use cases by impact, time-to-value, control requirements, and data readiness.
- Build once, scale often: Centralize common services (model gateways, PII redaction, logging, prompt libraries) to prevent tool sprawl and shadow AI.
- Own risk together: Finance, IT, security, and internal audit co-design controls and acceptance tests. No handoffs without signoff.
- Measure ROI like a CFO: Track savings, cycle-time reduction, error rates, forecast accuracy, and control outcomes-not just "adoption."
Shared principles to align on
- Data minimization by default; sensitive data never leaves approved environments.
- Audit-ready logging: prompts, outputs, training data lineage, and change history.
- Model lifecycle: approval, monitoring, fallback, and retirement criteria.
- Vendor governance: legal, security, SOC reports, exit clauses, and cost guardrails.
High-return use cases finance can lead now
- Close and reporting: variance explanations, narrative reporting drafts, reconciliations, anomaly detection.
- FP&A: forecast "copilots," driver-based planning, scenario generation, sensitivity analysis.
- Working capital: cash application, AP/AR matching, dispute resolution classification, supplier inquiries.
- Compliance: SOX testing assistance, policy mapping, evidence collection, control rationalization.
- Commercial finance: pricing and discount analysis, contract clause extraction, revenue recognition pre-checks.
If you need a starting point for vendor shortlists, this curated list of AI tools for finance can help teams compare options faster: AI tools for finance.
Governance and assurance without slowing momentum
You can tighten controls and ship faster by designing compliance into the workflow. The survey makes it clear: executives care deeply about SOX, ICFR, and SOC reporting-and CFOs feel that pressure most.
- SOX/ICFR: Tie each AI use case to a control objective and evidence artifact. Keep a control matrix for prompts, outputs, and model updates. For reference, see the Sarbanes-Oxley Act via the SEC: SOX (SEC).
- SOC reports: Require current SOC reports from AI vendors and cloud providers; log exceptions and compensating controls. A quick primer on SOC is here: SOC overview (ISACA).
- Model risk: Document data sources, bias checks, performance thresholds, and rollback plans. Monitor drift and incidents like you would key controls.
How to measure ROI in ways the C-suite trusts
- Efficiency: hours saved per cycle, close time reduction, invoice touchless rate, ticket deflection.
- Quality: error rate, rework, audit findings, exception volume.
- Speed to insight: time from question to analysis, scenario coverage, forecast accuracy delta.
- Cost: cloud and model spend per use case, vendor overlap eliminated, internal support cost.
- Adoption: weekly active users, repeat usage, satisfaction, and time to first value.
Report results monthly. Kill or fix low performers. Double down on winners. That discipline builds trust across the C-suite.
Funding and operating model that scales
- Product-based funding: Finance and IT co-fund multi-quarter roadmaps instead of scattered pilots.
- Central platform, federated delivery: Shared guardrails and services, with domain squads delivering use cases.
- Chargeback with transparency: Show unit economics per use case to keep demand honest.
Pitfalls to avoid
- Pilots that never graduate to production.
- Vendor sprawl and duplicative licenses.
- Unapproved data flows and weak logging.
- Manual review theater-no clear acceptance criteria.
- Letting either finance or tech "own" AI alone. Shared ownership is the point.
Bottom line
AI will not fix silos-leadership will. CFOs are in the best position to connect financial rigor with technical delivery. Align on principles, run one portfolio, and measure what matters.
Do that, and you move faster with fewer surprises-while staying audit-ready.
Your membership also unlocks:
