🎧 Listen to the full FEI Podcast episode »
The episode features Ryan Langenbrunner, Worldwide Leader for Planning Analytics with IBM’s Chief Data Office; Bill Primerano, Worldwide CTO for AI Analytics at IBM; and Michael Bernaiche, Director of AI Analytics Software at Lodestar Solutions, an IBM Gold Business Partner. All three started on the user side of financial planning before moving into technology, and each has lived through the spreadsheet-driven budget cycles many financial leaders are entering right now.
The conversation opens with a pointed question: How many of your executives could clearly articulate the organization’s strategic goals off the top of their heads? When it was posed to a room full of CFOs, only a handful raised their hands. If a budget is technically accurate but disconnected from where the company is headed, the panel argues, the organization doesn’t have a numbers problem. It has an alignment problem – and no amount of speed fixes that, not even with AI.
From Data Collector to Strategist
Langenbrunner began his IBM career as a financial analyst and later served as a worldwide forecast coordinator. He describes a cycle that will sound familiar to many listeners: data released in the morning, one hour to analyze it, and a 120-slide control book whose first 25 to 30 slides had to be ready before the first review meeting. “It was assembly,” he says. Reviews focused on whether numbers were right rather than what to do about them, and he left each one with more tasks than he came in with.
He recalls sprinting down the hall before an earnings call, pulling pages off the printer to assemble the book. “I went to school to be a business analyst, to break down a P&L and find things,” he says. “I spent more time gathering data, pulling it together and putting together PowerPoints.”
Today, after IBM implemented its own Planning Analytics technology and moved into modern data architectures, the picture looks very different. Leaders see results through dashboards as soon as data is available, a trusted source with stronger governance means less time questioning numbers, and in-depth discussions that once happened two or three weeks into the close cycle now happen on the fly as part of a daily close. “The ability to adjust and adapt is astronomically faster than what it was before,” Langenbrunner says.
Accuracy Means Little Without Timeliness
Primerano, who came up through the controller’s office, FP&A and acquisition strategy before joining IBM, says company size doesn’t change the core need. CFOs at organizations of every size worry that their teams are stuck in Excel, when what they need is decision responsiveness. He points to his time at Snapple, where oil prices were moving throughout the day and the business needed to model the impact quickly – a challenge that echoes today’s volatility around tariffs and energy costs.
Forecast accuracy remains the top KPI many CFOs set for their teams, he notes, but it isn’t enough on its own. At one organization, a new CFO mandated a single version of the budget. The team delivered exactly that: version 1.32. “We spent so much time working on it to get that forecast accuracy that we missed all of the things that were going on within the market and within our customers and our supply chain,” Primerano says. “What’s the point of having an accurate forecast that is already outdated?”
AI as an Accelerant – and Augmented Intelligence
Primerano compares AI to lighter fluid on a smoker: It makes things move much faster, which is exactly why it must be used responsibly. For finance, he argues, the goal isn’t artificial intelligence that makes decisions on its own. “It’s augmented intelligence,” he says – a system that processes the work and serves it up in a way people can understand, so they can make conscious decisions with a human in the loop.
He is skeptical of the push toward fully autonomous planning, likening it to a driverless car that navigates every turn once you enter an address. He has yet to meet a CFO willing to press an easy button and walk away without applying judgment, such as knowing that a key customer is shifting purchases into the next quarter. “We don’t do that in finance,” he says.
Who Should Govern AI Spending?
Many AI initiatives are being justified by headcount savings, Primerano says, while overlooking the costs of tokens, data movement and running large and small language models. CFOs may eventually find that expected savings simply shifted costs to the IT budget. That’s one reason he sees the office of the CIO increasingly rolling up into the CFO, with finance monitoring both sides of the equation when calculating AI ROI.
Langenbrunner says IBM already works this way. Its global AI leader and organization report into the CFO, and he compares the moment to the rise of SaaS, when new offerings promised the world and money flowed into business units. IBM responded by building business cases and checking in regularly: What are we spending? What are we getting back? Is it delivering at the rate and pace that was promised? AI spending is part of IBM’s overall CIO planning in Planning Analytics, much as cloud spending evolved from a single line item into dedicated planning cubes tracking capacity metrics. The panel expects many organizations to build planning models specifically around AI expenditure in the near future.
Alignment Before Speed
Bernaiche has worked with IBM Planning Analytics and Cognos Analytics for close to 30 years. Early on, he says, the focus was speed: build models fast and get data out. Over time, the missing ingredient became clear. “If you’re not aligned, all you’re doing is really speeding up broken processes and broken numbers,” he says. When executives aren’t aligned on the goals set with the board and CEO, faster tools only produce bad information more quickly – along with endless forecast versions and rework.
Primerano adds that many organizations still copy last year’s files, add a column or apply a growth factor to a 9+3 forecast, and assume the business model hasn’t changed. “The companies that are winning aren’t the ones with the biggest spreadsheets anymore,” he says. “It’s the ones that can model.” He cautions that AI, including tools such as Microsoft Copilot, is only as good as the data, governance and planning models behind it. Without trusted data, explainability and data lineage, the output is no better than growing last year by 4% and hoping the world doesn’t change.
The Cost of Falling Behind
The panel also addresses organizations that run planning technology on premise and upgrade every 18 months or so. Bernaiche notes that IBM now updates Planning Analytics roughly monthly, adding enhancements and built-in AI that make life easier for modelers and end users. “If you’re waiting 18 months to upgrade, five years ago you weren’t that far behind,” he says. “Today, you’re literally 18 months behind.”
He says moving to SaaS keeps organizations current and can limit audit and compliance risk, and he pushes back on the idea that enterprise planning tools are only for large companies. Even a team of five users can benefit, he says – especially on a Monday morning when new tariffs are announced and a rigid, cookie-cutter tool or spreadsheet makes rapid scenario changes painful.
Be Wary of the Shiny Demo
Langenbrunner has seen several internal groups at IBM pursue proofs of concept with vendors that promised fast, plug-and-play results. Each ran into the same brick wall: tools that look impressive at the top but are only as linear as the management system beneath them, creating silos and disconnecting from the rest of the organization. He credits multidimensional technology with avoiding that trap.
Bernaiche shares the story of a client that left for a plug-and-play tool, only to return one month before its budget launch because it couldn’t produce reports or customize the tool for its business. That client is now on SaaS. His advice: “If you haven’t come up with your requirements for a demo, don’t see the demo.”
In another case, a large Washington, D.C., law firm spent six months meeting with every vendor in the marketplace without ever writing a requirements document. Lodestar helped the firm build one, delivered a demo the next day and moved the firm, as Bernaiche puts it, “from uncertain to certain in three days.”
Primerano says evaluations too often rely on generic checklists of features nearly every planning tool offers, rather than asking whether a solution supports real business decisions and can be easily adopted. “Don’t buy for today’s requirements,” he says. “Instead, you need to think of what is tomorrow’s complexity.” He notes that COVID exposed how many organizations lacked a cash flow forecasting strategy, a reminder to plan beyond operating expenses and the annual budget.
Start Building Your Requirements Now
For financial leaders already working through this year’s budget, Bernaiche offers a simple first step: keep a notebook or whiteboard nearby and write down what you hate, what you love, what you want and what you need as you go. At the end of the cycle, review the list in a retrospective. “You’re starting to create your requirements as you’re doing the process,” he says. “It’s the best time to do it.”
Each panelist closes with one takeaway. Langenbrunner urges listeners to get started now, noting that planning is about unlocking insights from an organization’s most intelligent people, and that thinking which lives offline will be hard for any AI-era system to find. Primerano advises keeping the flexibility people value in Excel, “but stop asking Excel to be the database, the workflow system, audit trail, and enterprise planning engine.” Bernaiche returns to alignment: align internally so that you can confidently defend your numbers to your executive team and board.
This episode of the FEI Podcast is sponsored by Lodestar Solutions. Host Heather Cole is the founder and owner of Lodestar Solutions.
🎧 Listen to the full FEI Podcast episode »