The tech conference circuit is an exercise in collective hallucination. Walk the floors of VivaTech, and you are bombarded by a single, exhausting narrative: a hot European startup scales into a unicorn, bakes artificial intelligence into its enterprise software, and suddenly every CFO in the Fortune 500 becomes a strategic genius.
The latest poster child for this fantasy is Pigment. The business planning platform has earned rave reviews for its slick interface and its promise to replace old-school spreadsheets with real-time data and automated, intelligent modeling. The consensus among the tech press and corporate cheerleaders is lazy and unanimous: Pigment plus AI equals the death of spreadsheet errors and the dawn of frictionless corporate forecasting.
It is a beautiful story. It is also entirely wrong.
The enterprise software market is buying into a massive delusion. The problem with corporate planning has never been a lack of compute power, nor has it been the structural limitations of standard grid software. The problem is data politics and human incentives. By automating and speeding up the forecasting process, platforms like Pigment do not fix corporate strategy. They merely allow executives to generate flawed, politically motivated projections at a scale and velocity never before seen in enterprise history.
The Spreadsheet Myth: Why the Grid is Not the Enemy
For a decade, enterprise software vendors have used the exact same boogeyman to sell their products: Microsoft Excel. They point to the infamous London Whale trading debacle of 2012—where a copy-and-paste error in an Excel spreadsheet contributed to a $6.2 billion trading loss for JPMorgan Chase—as definitive proof that traditional spreadsheets are too dangerous for modern business.
This is a fundamental misunderstanding of why corporate models fail.
I have spent fifteen years inside boardroom discussions, watching multi-billion-dollar budgets get allocated based on financial models. I can tell you exactly how the sausage is made. The London Whale did not happen because Excel is bad software. It happened because the human beings operating the model were under immense pressure to justify risky positions, leading to a culture that ignored basic validation steps.
Excel remains the most successful software ever built because it is a blank canvas. It reflects the raw truth of an organization's logic, for better or worse.
When a modern planning platform claims to eliminate spreadsheet risk by locking down data pipelines and automating model building via natural language processing, it misses the point. It treats financial planning as a math problem. In reality, financial planning is a game of corporate theater.
Imagine a scenario where a regional vice president needs to hit a aggressive revenue target to secure their annual bonus. In an Excel sheet, they tweak the growth rate variable from 5% to 8%. It is obvious, transparent, and easily audited by anyone who traces the cell references.
When you move that same process into a complex enterprise planning engine driven by automated data feeds and conversational interfaces, that manipulation does not disappear. It just goes underground. The executive simply changes the underlying assumptions fed into the system or selects a different historical baseline that justifies their preferred outcome. The system then generates a beautifully polished, dashboard-ready report that looks completely objective but is just as compromised as the old spreadsheet.
Except now, because it has the stamp of a unicorn platform and a proprietary model behind it, nobody questions it. The software sanitizes bad data and gives it the illusion of scientific certainty.
The Speed Trap: Fast Decisions Are Often Disastrous
The core value proposition of the new wave of planning software is velocity. The marketing tells you that instead of waiting weeks for the finance team to consolidate numbers at the end of the quarter, you can query your data instantly. You can run hundreds of parallel scenarios in minutes.
But who actually benefits from a real-time, automated corporate forecast?
In the real world of industrial manufacturing, supply chain logistics, and long-term capital expenditure, hyper-velocity planning is a recipe for operational whiplash. Consider the basic mechanics of corporate decision-making. If a global consumer goods company shifts its five-year raw material procurement strategy every time a machine learning algorithm detects a two-week blurb in commodities pricing, the operational chaos would bankrupt the firm.
True strategic planning requires friction. It requires a slow, deliberate cadence where human beings are forced to sit in a room, defend their assumptions, and debate the structural realities of their industry.
When you automate the generation of these models, you remove the intellectual sweat equity required to understand them. When a junior financial analyst spends three days building a forecasting model from scratch in a spreadsheet, they understand every single moving part, every mathematical relationship, and every structural vulnerability of that model. They can tell you exactly where it will break.
If that same analyst simply types a prompt into a corporate planning tool to generate a three-year workforce optimization plan, they are driving a car with the hood welded shut. They cannot explain the variance because they did not build the logic; they merely accepted the output.
Dismantling the VivaTech Hype: The Real Cost of Enterprise AI
The broader issue on display at events like VivaTech is the complete lack of economic realism regarding enterprise AI integrations. Let's peel back the layers of how modern planning platforms actually operate under the hood.
Most corporate planning tools are essentially relational databases with a highly polished user interface layer. When they introduce conversational capabilities—allowing an executive to ask, "What happens to our margins if shipping costs double?"—they are not using an AI that understands international maritime logistics. They are using a large language model to translate natural language into a database query, running that query through a standard mathematical formula, and translating the result back into prose.
This introduces a massive, hidden layer of operational risk:
- Semantic Drift: A language model might interpret "gross margin" differently than your specific accounting department defines it, leading to subtle, undetected discrepancies in structural calculations.
- The Black Box Dilemma: Traditional enterprise planning software operates on deterministic logic. If input $A$ changes, output $B$ changes by a predictable factor. Introducing probabilistic elements into the data interpretation layer means two different executives could query the system with slightly different phrasing and receive conflicting strategic vectors.
- The Maintenance Debt: Training and fine-tuning models to understand the highly specific, messy jargon of a legacy multinational corporation requires an astronomical amount of clean, structured historical data—the one asset almost no Fortune 500 company actually possesses.
The tech industry loves to celebrate the valuation milestones of companies that promise to solve these problems. But a valuation is not a validation of a product's utility; it is a reflection of venture capital momentum.
The Hard Truth for CFOs: The Solution is People, Not Platforms
If you want to fix your organization's forecasting, stop writing eight-figure checks to software vendors promising automated salvation. The modern enterprise does not need a more sophisticated oracle; it needs a more honest culture.
True financial resilience is built on three unglamorous, non-technical pillars that no startup can package into an application interface:
- Radical Incentive Realignment: If your sales managers are compensated solely on volume, their data inputs into any planning system will always be structurally biased toward over-optimism. No amount of automated validation will correct an input that is intentionally designed to protect a bonus.
- Forced Simplification: If your financial models are so complex that they require a specialized platform just to run a basic sensitivity analysis, your models are too complicated to be useful. The best strategic decisions are made on models that can be sketched on a whiteboard and understood by a plant manager, not just a data scientist.
- Strict Intellectual Accountability: Implement a mandatory post-mortem process where historical forecasts are systematically compared against actual outcomes every six months. Force the teams who created the models to explain exactly why their assumptions failed.
The contrarian approach to modern enterprise tech is simple: treat every promise of automated efficiency with extreme skepticism. The software vendors who dominate major tech festivals are selling an escape hatch from the brutal, messy reality of human management. They want you to believe that if you just buy their platform, your data will become clean, your strategy will become clear, and your future will become predictable.
It won't.
Your organization will still be run by conflicted humans using dirty data to navigate an chaotic world. The only difference is that if you buy into the hype, you will be doing it with a significantly higher software licensing bill.
Stop trying to outsource your strategic thinking to automated platforms. Dust off your basic financial principles, look at your raw operational metrics, and force your leadership team to do the heavy intellectual lifting themselves. The grid isn't breaking your business. Your refusal to face reality is.
