AI-Ready Finance: The Practitioner's Guide
18 May 2026
Most AI-in-finance content is written by vendors. They have something to sell, so the frame is always “buy this, deploy that, here is the case study.” The honest version is different. Most finance functions are not ready for AI. The work to make them ready is the same work most of them have been deferring for years. Once you understand that, the order of operations becomes obvious.
This is the framework I use when assessing a finance function for AI readiness. Five dimensions, a self-assessment scorecard at the end, and a 90-day action plan that focuses on the right things in the right order.
The five readiness dimensions
There are five things that need to be true before AI gives a finance function meaningful leverage. Get all five right and the AI work becomes obvious. Skip any of them and AI produces wrong numbers faster than your team used to.
1. Data quality
The single most ignored prerequisite. AI is statistical pattern-matching on the data it is given. If that data is incomplete, inconsistent, duplicated across systems, or carrying years of bad coding decisions, the AI is going to confidently produce wrong answers.
Test it by asking a finance team a basic question. Total revenue by customer for the last twelve months. Margin by product line. Working capital trend by entity. If the answer involves three spreadsheets, a manual reconciliation, and a sentence like “give me an hour,” your data is not AI-ready. You do not need an AI vendor. You need data hygiene.
2. Process governance
Documented process, owned at the executive layer, with clear inputs and clear outputs. Most finance functions run on tribal knowledge. The treasurer who knows where the FX hedges live. The accounts payable lead who knows which approvers actually approve and which forward emails to their assistant. AI cannot operate inside undocumented process. It either guesses or fails.
The test: can a competent finance professional join your team and execute month-end correctly within two cycles, working only from your documented procedures? If not, your governance is informal. Informal governance does not survive contact with AI tooling.
3. Systems architecture
AI works best where data flows cleanly between systems. ERP to BI to consolidation to reporting. Mid-market finance functions are typically held together by Excel and the goodwill of whoever has been there longest. Excel is not the enemy. Excel as the system of record is the enemy, because no AI tool can be the source of truth when the source of truth is a workbook on someone’s desktop.
Map your finance systems. If the map has Excel sitting in the middle of it doing the work an ERP should be doing, AI is the wrong project. Systems integration is the right project.
4. Executive sponsorship
AI in finance is a change management programme that happens to involve software. The technology is the easy part. The harder part is that you are asking finance teams to trust outputs they did not produce by hand, to adopt new ways of working, and to retire workflows they have spent careers perfecting.
Without genuine executive sponsorship, that change does not happen. Without the CFO and the CEO behind it, the team reverts to the old workflow within a quarter. AI readiness is a board-level commitment, not a finance-team initiative.
5. Team capability
Finance teams need to understand what AI is and what it is not. They need to know enough to spot wrong answers, to question outputs, to use the tools well rather than passively. This is not training to become AI engineers. This is training to be informed practitioners.
The 40-staff AI literacy programme I delivered for a mid-market FMCG group was built around three questions. What is this technology actually doing. Where is it useful. Where will it fail. Those three questions, answered properly, get a finance team from suspicious to capable in six to eight weeks.
The self-assessment scorecard
Score each dimension 1 to 5. Be honest. The scorecard is for you, not for the vendor pitching you.
| Dimension | 1 — Not ready | 3 — Workable | 5 — AI-ready |
|---|---|---|---|
| Data quality | Multiple sources, manual reconciliation, frequent disputes about the numbers | One source of truth, occasional cleanup work | Clean, governed, single source, audit-ready |
| Process governance | Tribal knowledge, undocumented exceptions, key-person dependencies | Documented at a high level, gaps in detail | Fully documented, executive-owned, regularly reviewed |
| Systems architecture | Excel as the source of truth, manual data movement | ERP exists, but parallel spreadsheets do real work | Integrated systems, ERP is authoritative, data flows automatically |
| Executive sponsorship | ”We should look at AI” from a director, no budget, no mandate | CFO interested, no board commitment | Board mandate, CFO accountable, funded transformation programme |
| Team capability | No AI literacy, team is fearful or dismissive | Some team members curious, no structured learning | Whole team trained, can evaluate AI outputs critically, can use tools well |
Add the scores. Twenty-five is fully ready. Below fifteen, you are not ready, and the work is the prerequisites. The AI deployment is not the project. The prerequisites are the project.
The 90-day action plan
For finance functions scoring below twenty, here is the ninety-day plan. It is not an AI deployment plan. It is the prerequisites plan, which is the real AI readiness plan.
Days 1 to 30. Diagnose. Run the scorecard with your senior finance team. Map your systems. Audit your data quality on three high-stakes dimensions: customer, product, vendor. Document one full month-end process end-to-end as it actually runs, not as it is meant to run. Identify the three biggest dependencies on individuals rather than systems.
Days 31 to 60. Decide. Take the diagnostic output to the executive team. Propose the prerequisite programme, scoped honestly. Get budget and mandate. Decide whether the prerequisite work is finance-led, IT-led, or jointly led. Hire or contract the capability you need. Set the next six months of milestones in writing.
Days 61 to 90. Start the right work. Pick the highest-leverage prerequisite and start. Usually this is either data quality on a single critical dimension or systems integration to retire one specific spreadsheet. Resist the temptation to start AI vendor evaluations during this period. Vendor evaluations are a step seven activity, not a step one activity.
What this guide is not
It is not a prediction about which AI tools will win in the mid-market. I have opinions, but they will be wrong within eighteen months.
It is not a pitch to engage me. The work in this guide is the work most mid-market finance functions need to do whether they are pursuing AI or not. It is the finance transformation work.
It is not a promise that AI will transform your finance function. The honest version is that AI gives a well-run finance function genuine leverage. It does very little for a poorly-run one, beyond surfacing the problems faster.
For senior search firms and hiring decision-makers evaluating finance leadership candidates on AI readiness, the same five dimensions apply to your assessment criteria. Ask the candidate about their work on the prerequisites, not their AI vendor preferences.
Get in touch if you want to discuss how this framework applies to a specific brief.