AI training for CFOs and finance teams: a practical 90-day roadmap

Days 1 to 15: policy and baseline

• Agree which AI tools are approved and what data may be used in each.
• Pick 10 candidate processes, such as variance commentary, reconciliations and invoice extraction.
• Measure the current effort for each process in hours per month.
• Name an owner for AI adoption in the finance team.

Days 16 to 45: train and launch

Run hands-on training built on the team's own processes. Each participant should leave having completed real tasks, not watched demos.

Launch five use cases with the best ratio of value to effort. Month-end variance commentary, reconciliation break classification and MIS narrative are common first choices because the review step is clear.

Days 46 to 75: measure and fix

• Track hours saved and error rates against the baseline.
• Collect prompts that work into a shared prompt library.
• Fix the use cases that are not working, or drop them.
• Check that review and documentation steps are being followed.

Days 76 to 90: scale

Report results to the CFO with numbers, not anecdotes. Choose the next five use cases, including at least one that needs automation or analytics, and plan training for adjacent teams such as FP&A, tax or audit.

Common mistakes

• Buying tools before training people
• Starting with the hardest process
• No review step for AI-generated numbers
• Not measuring the baseline, so results cannot be shown