Dear colleagues,

AI can now produce a planning recommendation. The harder question is whether FP&A can explain why it should be trusted.

The 2026 FP&A Trends Survey shows how unevenly that responsibility is held: 41% of high-performing teams have formal AI policies in place, against just 12% of teams that are coping or struggling. Stronger teams are far less willing to leave AI accountability informal. That gap matters because AI is moving quickly from analysis towards recommendation. It can surface patterns, draft scenarios and suggest actions, but the harder question remains: who can challenge the answer, change it and stand behind the final recommendation?

This issue looks at that question. Our featured guide, How to Build AI-Driven Financial Planning Recommendations Your CFO Can Trust, sets out six steps for turning AI outputs into recommendations finance can defend, with clear traceability from source data and assumptions to model logic, finance review and audit trail. The articles take the discussion further: where automation should stop, how definitions and ownership shape AI output, and what FP&A must still bring that the model cannot.

A human checkpoint is not governance. Governance begins when challenge, override and accountability are designed into the process.

Warm regards,

Larysa Melnychuk

CEO and Founder at FP&A Trends Group

In finance, “the AI said so” is not an answer. A recommendation carries a chain of ownership: whose data, which assumptions, what logic, who reviewed it. This guide sets out six steps that keep that chain intact, from source data to a recommendation finance can defend, and makes the point that review only counts when it can change the answer, not merely approve it.

Read the six-step approach → 

Finance has a strange job: it is asked to explain decisions it had no part in making. When Sales, Product or Data teams embed the definitions, FP&A is left to reconcile and defend numbers it never shaped. The argument here is that ownership has to move upstream, into the logic before the model runs, not the audit after the forecast lands.

Read why ownership must move upstream → 

An AI agent managing working capital spots late payments and quietly tightens a customer's credit terms. Logical, and possibly wrong. Capability without accountability, as the piece puts it, is risk wearing a suit. It sets a direct operational test for how much autonomy any workflow should carry: what did the agent do, why, who authorised it, and what happened next?

Test the boundaries of autonomy → 

In 2021, Zillow's home-buying algorithm kept bidding as the market turned, on resale assumptions that no longer held. No one was tasked to challenge or stop it, and the write-downs passed 500 million dollars. This article names the missing role, the FP&A Agent Manager: a domain expert who tests an agent's assumptions against commercial reality and owns the controls before failure reaches the financial statements.

Read how the role should be designed → 

Our webinars are free and educational, and help thousands of FP&A teams around the world learn together. Join live when you can, or register and we will send you the recording to watch in your own time.

The upcoming sessions carry this issue's question forward from two sides: how Finance builds the trust and capability to adopt AI, and how it retains ownership as faster analysis moves into decisions.

FP&A transformation often stalls when adoption is treated as an afterthought. The discussion focuses on how leaders build trust, capability and behavioural change into new ways of working.

Register now → 

Faster analysis does not automatically produce better decisions. The webinar examines how Finance turns AI-enabled insight into action while keeping clear ownership of the call.

Register now →