Skip to main content
logo top bar

The Online Resource for Modern FP&A Professionals

Main menu

  • Home
  • FP&A Insights
    • FP&A Articles
    • FP&A Publications
    • FP&A Trends Digest
    • Short Videos
    • Our Contributors
  • FP&A Events
    • International FP&A Board
    • FP&A Trends Webinars
    • Digital FP&A Circles
  • AI FP&A Committee
    • Introduction
    • Members
    • Resources
    • Meetings
  • FP&A Maturity Model
  • About Us
    • Company Policy
    • Privacy Policy
    • Editorial Guidelines
    • Our Ambassadors
    • Our Sponsors & Partners
    • Contact Us
  • Sign up
  • Login
image banner
AI Is Not "Other People" – Part 2 of 4
November 17, 2026

By Kevin Phillips, CEO at IDU

FP&A Tags
AI in FP&A
Collaborative FP&A
Planning and Budgeting
FP&A Business Partnering

Kevin-Phillips-AI-Is-Not-Other-People

This is the second article in a series for smaller finance teams putting AI to work in their planning. In the first, I made the case that the most dangerous AI is the one that agrees with you. I also mentioned that there are different flavours of AI and that which one you choose impacts your outcomes. This article is about one of those consequences, and it starts with the question I left hanging last time.

 If the machine will not challenge your assumptions, who will? The answer, of course, is other people. But this discussion, debate, and, probably, disagreement is what AI is quietly replacing as we strive for speed and efficiency.

Think about how a budget used to come together, at least when it was done well. You involved people. The operations manager who could see a supply problem coming before it showed up in any number. The sales lead who knew the pipeline was softer than the spreadsheet suggested. These people contributed on-the-ground information that AI, with its reliance on historical patterns, can never have access to. Plus, and most importantly, they were willing to (and often determined to) argue with you.

Spending less time debating? This is a warning light!

Today, in our race towards speed, agility and productivity, the temptation is to skip all of that and ask AI instead, because it is faster, it always answers, and it is highly agreeable. People, on the other hand, don’t like your spreadsheet, don’t understand financial jargon but don’t want to admit it, and postpone getting their figures back to you for as long as humanly possible.

Here is the trouble with consulting AI instead of other people. You miss the demand shift the sales team saw coming, so your revenue forecast is built on a number that was already out of date. You budget for capacity the operations manager knew did not exist. You commit to targets that the people responsible for hitting them had no hand in setting, and don’t think they can reach. None of this shows up in the plan itself, which looks as clean, confident, and delivered on time. It shows up a quarter or two later, in the variances nobody can explain and the reforecast you did not think you would need.

Efficiency Dressed Up as Collaboration

This is where the AI you choose matters. If you choose a tool that sits at the centre and produces the whole plan for you, it might feel like you have consulted widely, because the answer comes back so quickly and so completely. But this is efficiency dressed up as collaboration. A different type of AI, one built to bring the people around the plan in rather than shut them out, does the opposite.

Give Every Assumption an Owner

There is a growing view in this community that planning should be owned not by finance alone, but by the business as a whole. I agree with it, and I want to add the practical part, and explore how you make that happen.

To drive distributed assumption ownership in your organisation, you need to start by finding the handful of assumptions that drive most of your budget, and give each one a named owner who sits close to what it describes. Marketing owns the demand assumptions. Operations owns the efficiency assumptions. Treasury owns the currency view, and so on. Finance stops dictating every input and starts orchestrating the whole picture. These distributed owners carry information the model does not have, and because they have skin in the game, they will challenge you in ways AI can’t. A person who owns the assumption owns the outcome, and a person who owns the outcome will argue about the inputs. That argument is what keeps the plan honest.

That doesn’t mean abandoning AI though. The right AI can support this by carrying the workload, and leaving the ownership where it belongs, with the people who have to deliver the numbers.

Think back to the accountant I described in the first article, the one who built a clever in-house AI tool but did not fully trust its numbers. Picture how that plays out. The tool confidently forecasts raw material costs holding flat next year, drawing on three years of history that say exactly that. The budget looks clean. But the operations manager, who actually talks to suppliers, knows a major contract is up for renewal and the price is about to increase. If the operations manager owns that assumption, the number gets challenged before it reaches the board. If the model owns it, nobody in the room knows enough to argue, and the miss only shows up two quarters later. The tool was not wrong to use it. The accountant was wrong to trust it alone.

Keep the Argument

As I said, AI is still useful for the heavy lifting, the collation, the modelling, the first draft of a forecast that would once have taken you a week. But, according to the 2026 FP&A Trends Survey, FP&A still spends 47% of its time on activities like data collection and validation. Imagine the improved insights, analysis, and decision-making that could happen if this workload were moved to AI.

Use AI for the workload, and keep the debate for the people. That part was never meant for the machine. AI doesn’t care whether the plan is right, and it can’t sit in a room and disagree with you because it has seen something you haven't. The teams that get maximum benefit from using AI will not be the ones that handed the most over to AI. They will be the ones that used AI to carve out the time to have the important strategic conversations.

Before Your Next Planning Cycle

  • Name the owner of your five most critical planning assumptions. If the same name sits against all five, you have found your problem.

  • Before the next forecast, ask one coalface manager what the model does not know, and write down what comes back.

  • Pick one conversation you deferred to AI, and give it back to a person.

The full text is available for registered users. Please register to view the rest of the article.
  • Log In or Register

Related articles

Adinath-Kadam-AI-FPA-Line-Between
AI in FP&A: The Line Between Automation and Judgement
June 4, 2026

AI improves FP&A efficiency, but real value still depends on human judgment, business context, and accountability.

Read more
Andreas-Seufert-AI-FPA
AI in FP&A: From Analytical Signals to Management Decisions — Part 1 of 7
July 2, 2026

AI in FP&A creates value when earlier analytical signals are connected to governance, ownership, escalation, and...

Read more
Vignesh-Dumonceau-Agentic-Planning
From Excel-Based Drivers to Agentic Planning: A CFO Perspective
August 27, 2026

Agentic Planning helps CFOs move beyond Excel-based driver models by combining AI-enabled speed with stronger FP&A...

Read more
Tejas-Parikh-Leaders-Dont-Trust-Numbers
Why Senior Finance Leaders Still Don't Trust Their Own Numbers
October 22, 2026

Why finance leaders still recheck their numbers, and four practical ways to rebuild FP&A data trust...

Read more
Christian-Martinez-Teaching-200-Finance-Leaders-AI
What I Learned After Teaching 200+ Finance Leaders AI and Python for FP&A
October 27, 2026

AI makes FP&A analysis faster, but financial judgement and validation determine whether the output can actually...

Read more
Kevin-Phillips-The-Most-Dangerous-AI
The Most Dangerous AI Is the One That Agrees With You – Part 1 of 4
November 5, 2026

AI sycophancy can reinforce flawed assumptions and distort forecasts. Learn how FP&A teams can challenge AI...

Read more
+

Subscribe to
FP&A Trends Digest

We will regularly update you on the latest trends and developments in FP&A. Take the opportunity to have articles written by finance thought leaders delivered directly to your inbox; watch compelling webinars; connect with like-minded professionals; and become a part of our global community.

Create new account

Future Meetings

  • Digital
  • In person
The Digital APAC FP&A Circle
The Digital APAC FP&A Circle: Accelerating Continuous Planning in the Age of AI
Accelerating Continuous Planning in the Age of AI

November 18, 2026

The FP&A Trends Webinar
The FP&A Trends Webinar. Inside the Crystal Ball: Transformative Trends in FP&A in 2027 and Beyond
Inside the Crystal Ball: Transformative Trends in FP&A in 2027 and Beyond

December 9, 2026

Pagination

  • Previous
  • November 2026
  • Next
Su Mo Tu We Th Fr Sa
1
2
3
4
5
6
7
Leading FP&A Change in the AI Era
 
From AI Efficiency to Finance Value: The FP&A Value Architect in Action
 
Leading FP&A Change in the AI Era
 
Leading FP&A Change in the AI Era
 
8
9
10
11
12
13
14
Leading FP&A Change in the AI Era
 
15
16
17
18
19
20
21
Accelerating Continuous Planning in the Age of AI
 
22
23
24
25
26
27
28
29
30
1
2
3
4
5
 
 
 
 
 
All events for the year

Homepage – Acme Corp

Visit our register pageemailVisit our LinkedIn pagelinkedinVisit our Twitter profiletwiterWatch our YouTube channelyoutube Visit our register pagefp&a digest

A leading international think tank dedicated to advancing the practice of Financial Planning and Analysis (FP&A). Its mission is to shape the future of the profession by exploring emerging trends and sharing best practices through research, publications, events, education, and advisory work.

Foot menu

  • FP&A Articles
  • FP&A Publications
  • FP&A Trends Digest
  • International FP&A Board
  • FP&A Trends Webinars
  • AI FP&A Committee

© 2015 - 2026, FP&A Trends Group. All rights reserved.

0