A company should now expect FP&A to move from “give me a BVA report” to being the team that builds systems and brings in external context.
Instead of: The margin dropped because costs were up.
The expectation becomes: The margin dropped because we lost three contracts to X, Y, and Z, who lowered prices from $10K to $7K, and we had to offer three customers 50% discounts to keep the business.
Instead of: We are over budget because marketing spend was higher.
The expectation becomes: Marketing spend was higher because three LinkedIn campaigns were expected to reach 10,000 target ICP customers, but only reached 5,000. We then had to supplement with print ads, which were 50% more expensive, to generate enough leads for the sales team.
Instead of: What are the upside, downside, and base case scenarios?
It becomes: We are moving forward with X, Y, Z operational actions. Each depends on execution by a specific date. Here is the scenario range, here are the leading indicators, and the teams responsible are already working against it.
Instead of: Headcount is above plan.
The expectation becomes: Headcount is above plan because we had to hire three contractors in customer success to support four stalled implementations. Those contracts were worth $100K a month, and the new hires will cost $50K to get the work done. It was a high-ROI decision.
Now, we have spent a lot of time talking about how the current crop of tools can use AI to take us from the five circles of data hell to full planning orchestration. But most teams that try to make this pivot are finding that this results in: one team starts coding tools in Firebase, another starts loading the full G/L into a Google chatbot, and then management starts asking whether any of it actually works. (Cue AI overhyped LinkedIn post.) The inevitable backlash follows takes FP&A back to the ; previous way of working.
So let’s take a step back and look at how FP&A work will really change. (No, not vibe coding.)
People are being conditioned to expect that they can ask a question and get an answer in seconds. In that world, the typical FP&A model of “ask a question, build a dozen reports, embed one finance leader per team” stops making sense. If answers are at people’s fingertips, they do not need to wait on finance just to get basic information.
Claude Code and MCP make integrations dramatically easier. What used to take months of API work can now be done in hours by someone with basic technical knowledge. That opens up an entirely different range of tools and workflows.
What does that lead to?
An FP&A team that builds a system where anyone, anywhere in the business can access the data they need to make decisions without routing every question back through finance.
So what does that change the FP&A role into?
It shifts the role toward combining internal data with outside context and turning that into real decision support. That changes the function entirely.
That is a very different job. It is not just reporting the numbers. It is explaining what happened, why it happened, what is likely to happen next, and what the business should do about it.
So how do you build toward that system now?
Build on standard tools. Encourage AI experimentation absolutely. It is useful. But if you want something to go live, do not waste time recreating Slack or building a one-off internal tool that nobody will maintain. Build on platforms, infrastructure, and open-source tools that save time and can actually scale.
Evaluate your current toolset. You do not want your data locked away or flattened into something unusable. You need all of your data in one place — yes, including detailed budgets, assumptions, and dimensional history, not just a flat summary table. That is what lets you connect it to the rest of your tooling and actually use it properly.
Create a create-test-deploy strategy. One person in FP&A should own the final tool release. You can move incredibly fast now, but speed is not the same thing as permanence. The person overseeing this needs to understand the long-term architecture and the broader company vision, not just the tool being launched this week. We may see the rise of a financial engineer.
Consider the user. Nobody wants to run 20 different vibe-coded Python tools just to do their job. They want to access the right data and collaborate with their team in one place, ideally inside a tool they already use every day.
Over the last several articles, we’ve covered how to build this infrastructure from scratch: the data foundation, the models, the automated insights and foresight, and the orchestration layer that ties it all together.
Now try a simple exercise. Spend a few hours imagining the detail and information you would want to make better decisions: competitor pricing, supply chain options, market penetration, product analytics, and etc. Spend half of that time thinking of creative ways to get that information. That is your vision. And, that is what top CFOs and FP&A leaders are building towards now. Quickly.
