Thank you! Your submission has been received!
Oops! Something went wrong while submitting the form.
If a delivery partner tells you they have adopted AI, the fair question is: where did the efficiency go?
AI should make a partner faster and more accurate. As a public-sector buyer you are entitled to see that as better value for money.
This article gives you the questions to ask, so you can tell a partner using AI for real from one talking about it.
Private companies want value for money as much as anyone.
A private buyer who overpays answers to management or shareholders, and they answer for the result. If the project lands and the numbers work, nobody revisits how the supplier was chosen.
Public sector runs on process. You answer for how the decision was made, and you answer to people who were not in the room. A good outcome does not excuse a procurement you cannot evidence.
That is why a partner's unverifiable AI claims become your problem rather than theirs.
If you are running IBM Planning Analytics, the platform now includes an AI assistant, licensed per user, letting a finance team query data in plain language and run scenarios without waiting on a specialist.
That is a product capability you buy from IBM. Whether your delivery partner has changed how they work is a separate question, and the one this article is about.
You want a partner who uses AI. You do not want one who makes everything AI.
A partner ignoring it altogether is a red flag. Real ground is being covered in finance workflows now: reconciliations, anomaly detection, forecasting, invoice validation, variance explanation, report drafting. Working the slow manual way means charging you for effort that could be more reliable and less error-prone.
The other direction is a problem too. When every step has a bot and every process has an agent, the partner has usually lost sight of the problem they were hired to solve.
That one is harder to spot, because it looks like progress. The pitch gets more impressive while the delivery and the price stay where they were.
A partner leading with the technology will show you what the tools can do in the abstract, without connecting it to your work, your accuracy requirements, or your price.
A partner leading with the work will tell you which parts are now faster, which stay human-checked, and how that shows up in the engagement.
The tell is specificity. A generic answer could be lifted into any pitch for any client. A specific one is uncomfortable to fake, because it commits the partner to claims you can test.
These are fair, they are answerable, and a credible partner will welcome them.
You are asking about method, not capability.
A weak answer describes what AI can do in finance generally. A strong one describes how this partner's own process differs from two years ago: which steps are automated, where information is validated faster, how the engagement has changed shape.
Push for specifics. Is the close cycle quicker? Is validation less manual and therefore less error-prone? Are variances surfacing earlier rather than caught late?
A partner who can point to where their delivery became more consistent is describing reduced risk to you, which is what you are obliged to value.
If the answer is "not much yet," that is more trustworthy than a sweeping one.
The question most partners would prefer you did not ask.
If a partner has genuinely become faster, something should have moved: either the price, or what you receive for it.
You are not squeezing them. You are testing whether their AI story and their pricing story agree.
Finance carries very little tolerance for error, and public money is spent under external scrutiny. The right answer is never "the AI handles it."
Mature use keeps a person accountable for judgement, materiality and sign-off, with the tools doing ingestion, pattern-finding and first drafts.
A partner who waves the question away, implying oversight is no longer needed, should lower your confidence.
Concrete, modest, and internally consistent. It names what changed, names what did not, and shows the commercials match the story.
It will not claim every workflow has been transformed at once. Sensible partners tackle the feasible processes first and approach high-risk tasks carefully.
And it will not be afraid to say "we don't use it there, and here's why."
You do not need to be technical to have this conversation. You need a partner who can explain, in plain language, where these tools help, where they do not, and what that means for what you pay.
Most partners worth hiring can do that. They have thought it through already, because their clients have started asking.
So ask. A partner who answers these four questions straight is showing you how they will handle the rest of the engagement.